Compare commits

..
Author SHA1 Message Date
ДмитрийandClaude Sonnet 5.5 55b0087455 feat(modules): choose Parallel or Global scaling progression in module settings
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-22 15:59:29 +05:00
Corfirean dc05b2cd3c Record reviewed Manager 0.7.1 release notes 2026-10-22 11:00:19 +05:00
Corfirean 2e3f36d2fd docs: keep public guides and screenshots focused on users (#139) 2026-10-22 08:12:13 +05:00
Corfirean 8f39896f0c Publish released 0.7 player guides and screenshot gallery on master (#138) 2026-10-22 08:04:59 +05:00
github-actions[bot] 103c829539 Update published Manager changelog 2026-10-22 02:53:40 +00:00
Corfirean 20fb2391ed Record every published Manager version in its own changelog (#137) 2026-10-22 07:49:15 +05:00
Corfirean 43de03eada Keep imported upstream notes separate and read historical BOM fragments (#136) 2026-10-22 00:14:25 +05:00
Corfirean cda58cdcd1 Fix collecting the next published Manager changelog from its Git tag (#135) 2026-10-22 00:04:32 +05:00
ДмитрийandClaude Sonnet 5.5 f6eccf849f Content Scaling module settings: creature health multipliers
Five float fields (open world, dungeons, raids, ordinary mobs, bosses; 0.1 to 10) in six languages, with save-time validation in coa-core. They appear once the
server's Content Scaling module carries the matching CoAContentScaling.Difficulty.Health.* settings.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-10-21 15:09:43 +05:00
Corfirean 19791edb94 Preserve fragment validation alongside release-note synchronization (#127) 2026-10-20 06:38:25 +05:00
Corfirean 4537b2a826 Restore all published Manager changelogs and keep release notes synchronized (#126) 2026-10-20 06:37:28 +05:00
Corfirean 4a88f696a0 Link agents to the shared release and artifact handoff runbook (#123) 2026-10-19 19:32:50 +05:00
Corfirean e3fbb2e4b6 Update older versions directly using database migration history (#112) 2026-10-16 16:56:32 +05:00
Corfirean a7b56d6aed Simplify updates to one recovery-backed migration pass (#111)
* Share migration planning and execution across update phases

* Apply updates in one recovery-backed pass without nested server rehearsals

* Preserve applied core SQL and share account migration history across realms
2026-10-16 13:39:43 +05:00
Corfirean cda7875a9a Merge pull request #110 from Corfirean/release/0.6.19
Release Manager 0.6.19
2026-10-16 07:27:31 +05:00
Дмитрий f53c7cb221 Release Manager 0.6.19 2026-10-16 07:27:12 +05:00
Corfirean ca6bcf0052 Merge pull request #109 from Corfirean/fix/lightweight-update-check-20261016
Avoid archive downloads during checks and preflight combined rehearsal space
2026-10-16 06:02:05 +05:00
Дмитрий d6a6384f9c Keep update previews local and preflight the complete rehearsal footprint 2026-10-16 05:54:25 +05:00
Corfirean 21ddc1e7d8 Merge pull request #108 from Corfirean/fix/update-bridge-20261016
Restore updates from older server versions through tested intermediate releases
2026-10-16 05:27:19 +05:00
Дмитрий 16396407ad Offer signed intermediate upgrades for older server installations 2026-10-16 05:20:12 +05:00
Corfirean 093dbfcadc Merge pull request #102 from Corfirean/fix/stabilize-update-recovery-20261014
Fix Repair recovery, shutdown and missing update errors
2026-10-15 21:43:26 +05:00
Дмитрий dc323df681 Release Manager 0.6.17 2026-10-15 21:36:00 +05:00
Дмитрий 46785f7978 Stop the complete database before packaging or finishing qualification 2026-10-14 19:31:33 +05:00
Дмитрий 97c024d95c Install current stable full images and qualify clean installation 2026-10-14 19:22:45 +05:00
Дмитрий 52b80232fa Install CoA Custom 1.7 overlays with complete legacy race assets 2026-10-14 18:57:16 +05:00
Дмитрий be2f252593 Show Overview update failures and cover copied launcher configuration 2026-10-14 08:26:19 +05:00
Дмитрий 24d593d787 Roll back late repair failures and verify signed repair recovery 2026-10-14 07:25:56 +05:00
Дмитрий 4a9760b864 Recover shutdown with damaged templates and MySQL state 2026-10-14 07:07:35 +05:00
Дмитрий 482bbfcdfd Fix repair completion and expose SQUID and rehearsal failures 2026-10-14 06:52:24 +05:00
Corfirean b45e2502ec Merge pull request #96 from Corfirean/release/0.6.16
Release Manager 0.6.16
2026-10-13 17:24:10 +05:00
Дмитрий 6a955bd670 Release CoA Server Manager 0.6.16 2026-10-13 17:12:18 +05:00
Corfirean d5f441de0c Merge pull request #93 from Corfirean/fix/long-rehearsal-paths-84
Fix long Windows SQL paths in update rehearsals
2026-10-13 07:24:38 +05:00
Дмитрий 4efc02e768 fix(paths): recognize long Windows rehearsal paths inside server 2026-10-13 07:18:03 +05:00
Corfirean be9641254a Merge pull request #92 from Corfirean/feat/client-races-1.6
feat(client): update custom races to verified 1.6 payloads
2026-10-13 07:14:35 +05:00
Дмитрий c18daac228 docs(changelog): record supported experimental bot race preset 2026-10-13 07:02:07 +05:00
Дмитрий 89c62abf80 fix(bots): align experimental race exclusions with upstream 1.6 2026-10-13 07:01:15 +05:00
Дмитрий a6471d79d0 feat(client): update verified custom race payloads to 1.6 2026-10-13 06:52:12 +05:00
Corfirean fd0eaafda2 Merge pull request #70 from lostmind84/feat/linux-b4-parity
feat: Docker server parity on Linux (realm online, game data folder, settings, sharing)
2026-10-13 06:50:56 +05:00
Corfirean fb1cc38773 Merge pull request #73 from LarsSeverson/fix/local-host-friends-mode
fix(friends): preserve local realm address
2026-10-13 06:42:46 +05:00
Corfirean 7a950f234b Merge pull request #91 from Corfirean/fix/export-maintenance-diagnostics
fix(diagnostics): log refused starts and export maintenance ownership
2026-10-13 06:42:42 +05:00
Corfirean 2f1df13b7a Merge pull request #90 from Corfirean/fix/operation-lock-contention
fix(startup): wait for update checks and identify active maintenance
2026-10-13 06:42:38 +05:00
Corfirean be705329c1 Merge pull request #89 from Corfirean/fix/show-installed-module-versions
feat(modules): display installed server revisions and client patch version
2026-10-13 06:42:34 +05:00
Corfirean 7e70409ca0 Merge pull request #88 from Corfirean/fix/client-patch-preservation
fix(client): preserve and repair custom race patch files safely
2026-10-13 06:42:08 +05:00
Corfirean 9e6a650d13 Merge pull request #87 from Corfirean/fix/startup-failure-service-logs
fix(diagnostics): show service errors behind failed update rehearsal
2026-10-13 06:42:04 +05:00
Corfirean 0398a5a6f9 Merge pull request #86 from Corfirean/fix/stock-bot-race-exclusions
fix(bots): keep the standard-races preset free of custom race IDs
2026-10-13 06:42:00 +05:00
Corfirean dd23e81dd4 Merge pull request #85 from Corfirean/fix/recheck-installed-release
fix(updates): allow checking the already installed release
2026-10-13 06:41:56 +05:00
Дмитрий ca2a555d7c fix(docker): request restart when remote client protocol changes 2026-10-13 06:41:16 +05:00
Дмитрий 8269a53db7 fix(diagnostics): record refused starts and export maintenance ownership 2026-10-13 06:26:46 +05:00
Дмитрий b97f86fc00 feat(modules): show installed revisions and client patch versions 2026-10-13 06:21:05 +05:00
Дмитрий 358f9f4f2b fix(bots): exclude every nonstandard race in the stock preset 2026-10-13 06:17:52 +05:00
Дмитрий 0c0d2db368 fix(startup): wait for checks and report operation lock owners 2026-10-13 06:17:17 +05:00
Дмитрий acd04fae2f fix(diagnostics): include service errors after launcher readiness 2026-10-13 06:10:56 +05:00
Дмитрий 106a532092 fix(updates): accept rechecking the installed target release 2026-10-13 05:59:38 +05:00
LarsSeverson 51ef88101e fix(friends): preserve local realm address 2026-10-11 11:35:24 -05:00
Fabien F. 4d8ca00e46 test: a Docker config built by a test has no database data directory 2026-10-11 08:57:43 +02:00
Fabien F. 36bfc6d31e docs: changelog entry for Docker server parity 2026-10-11 08:57:43 +02:00
Fabien F. 3fb4ee39c6 fix: the realm is not shown offline after a start on Docker
The auth server marks every realm offline when it starts, after the world server
cleared its own flag. Mark the active realm online once the auth server listens.
2026-10-11 08:57:43 +02:00
Fabien F. 0250b61c62 feat: a wrong game data folder is diagnosed, and it can be changed from Settings
The world refusing to start with the standard Data set (no CoA DBC) is now
reported as GameDataMismatch. A Settings card shows the folder in use and
changes it while the server is stopped, using the install checks.
2026-10-11 08:57:43 +02:00
Fabien F. dfd8a50975 fix: sharing with friends and the other screens work on a Docker server
Second part of the audit of every screen for a Docker server. What it found, and
what now happens:

- Play with friends. "Servers open" read the BindIP of the configuration files, which
  the containers force to 0.0.0.0, so it was green on a fresh server while who can
  connect is decided by the address the ports are published on (bindAddress in
  docker.json). Opening and closing now edit bindAddress (and set
  CoA.AllowRemoteClients), the open state is read from it, and enabling a mode no
  longer fails on netsh / PowerShell after it already rewrote the files. There is no
  firewall row on Linux (Docker publishes the ports itself); the Windows reply is
  unchanged.
- The "private services" check (Friends and Diagnostics) could never fail on Linux: it
  reads the Windows listener table. For a Docker server it is computed from how the
  ports are published: the database never, the console only on 127.0.0.1, the game
  servers on bindAddress.
- Console, Database tab: read a mysql log file that does not exist; it shows
  `docker logs` of the database container.
- The Overview showed a service that had stopped by itself as "Stopped"; it now says
  "Stopped unexpectedly".
- Settings: a Docker server that was added but never started had no
  worldserver.conf and the page failed; it is created from its template. A talent
  file path outside the Core folder is refused (the container cannot see it).
- Backups of configuration now include docker.json; verifying the files of an
  imported Docker server looked for worldserver.exe and reported it missing.
- Linux locations: the friend package and the diagnostics go to the desktop folder
  (from the user directories file) instead of /tmp, the problem report names the
  distribution, and the size advice for companions reads /proc/meminfo instead of
  assuming 8 GB.

Every Windows branch is the code that was there; the dispatch is the early return on
docker::is_docker (or the host flavor for the firewall).
2026-10-11 08:57:38 +02:00
Fabien F. 90c308615b fix: settings the Docker backend fixes are not offered, and the world restarts itself
Found by auditing every screen for a Docker server.

- Server settings: Ra.Enable, Ra.IP and Updates.EnableDatabases were on the page and
  written to worldserver.conf, but the containers get AC_RA_ENABLE, AC_RA_IP and
  AC_UPDATES_ENABLE_DATABASES, and the core reads those before the file: the change
  was accepted and had no effect, and the page showed the file's value while the
  container ran another. A Docker server no longer offers them (nor "restore
  recommended defaults" for them), and a change forced through is refused with the
  reason. The list is docker::FORCED_KEYS, and a test compares it to the variables
  the containers really get, so adding one without the other fails.
- `server restart` (and idlerestart) make the core exit with code 2 and expect a
  supervisor to start it again; the Docker world had none and showed "crashed". It now
  gets `--restart on-failure:3`, applied once the server is up (so a start that fails
  stays failed instead of looping); a stop or `server shutdown` (code 0) is never
  restarted.
2026-10-11 08:57:38 +02:00
127 changed files with 4486 additions and 1254 deletions

No files matched your search

+14
View File
@@ -0,0 +1,14 @@
name: changelog-fragments
# Checks the changelog.d/ fragment files with the shared rules (https://github.com/Corfirean/coa-server-build/blob/main/changelog/GUIDE.md).
on:
push:
paths: ['changelog.d/**']
pull_request:
paths: ['changelog.d/**']
permissions:
contents: read
jobs:
check:
uses: Corfirean/coa-server-build/.github/workflows/changelog-check.yml@main
+41 -8
View File
@@ -1,14 +1,47 @@
name: changelog
# Checks the changelog.d/ fragment files with the shared rules (https://github.com/Corfirean/coa-server-build/blob/main/changelog/GUIDE.md).
on:
push:
paths: ['changelog.d/**']
branches: [master]
paths: [tools/sync_history.py, changelog/history.json, .github/workflows/changelog.yml]
pull_request:
paths: ['changelog.d/**']
paths: [tools/sync_history.py, tools/test_sync_history.py, changelog/history.json, .github/workflows/changelog.yml]
release:
types: [published, prereleased, released]
workflow_run:
workflows: [release]
types: [completed]
workflow_dispatch:
permissions:
contents: read
contents: write
concurrency:
group: component-changelog
cancel-in-progress: false
jobs:
check:
uses: Corfirean/coa-server-build/.github/workflows/changelog-check.yml@main
test:
runs-on: ubuntu-latest
permissions:
contents: read
steps:
- uses: actions/checkout@v7
- run: python3 -m unittest discover -s tools -p test_sync_history.py
sync:
needs: test
if: github.event_name != 'pull_request' && (github.event_name != 'workflow_run' || github.event.workflow_run.conclusion == 'success')
runs-on: ubuntu-latest
env:
GH_TOKEN: ${{ github.token }}
steps:
- uses: actions/checkout@v7
with:
ref: ${{ github.event.repository.default_branch }}
fetch-depth: 0
- name: Synchronize this component from published immutable sources
run: python3 tools/sync_history.py --sync-release-notes
- name: Commit updated history
run: |
git config user.name 'github-actions[bot]'
git config user.email '41898282+github-actions[bot]@users.noreply.github.com'
git add CHANGELOG.md changelog/history.json
if git diff --cached --quiet; then exit 0; fi
git commit -m 'Update published Manager changelog'
git push origin HEAD:"${{ github.event.repository.default_branch }}"
+24 -2
View File
@@ -3,8 +3,7 @@
<!-- changelog:start (managed by coa-server-build/changelog/changelog.py sync-instructions) -->
## Changelog (shared by all CoA repositories)
Every CoA project (server fork, bots, Content Scaling, Manager, renderer) feeds **one** changelog that is posted to the
community. When you finish a change that a **player** or a **server owner** would notice (a new feature, a changed
Every CoA project (server fork, bots, Content Scaling, Manager, renderer) uses the same fragment format. **Manager and server release histories are separate**; each repository publishes only its own shipped component changes. When you finish a change that a **player** or a **server owner** would notice (a new feature, a changed
behaviour, a fix for something people ran into, a removed feature, a known issue), add **one small file to
`changelog.d/` in the same commit**. Do not add one for internal work (refactors, tests, CI, documentation).
@@ -29,3 +28,26 @@ Optional: one short paragraph in plain English for the reader (at most 300 chara
`python changelog.py new --area bots --type fixed --audience players --title "..."`.
- Full guide with examples: https://github.com/Corfirean/coa-server-build/blob/main/changelog/GUIDE.md
<!-- changelog:end -->
## Shared build and release workflow
Every published Manager version, including a clearly marked Pre-release, belongs in this repository's
CHANGELOG.md and changelog/history.json. Its GitHub release notes use the same reviewed record.
Promotion updates that entry without duplicating it; Manager and server histories remain separate.
Before preparing or publishing a Manager/server release, read the maintained
[release runbook](https://github.com/Corfirean/coa-server-build/blob/main/docs/server-release.md)
and [build agent instructions](https://github.com/Corfirean/coa-server-build/blob/main/AGENTS.md).
Use the local build checkout when available; otherwise retrieve the current files from GitHub.
Recover the exact prepared run, source commit, version, hashes and artifact expiry before rebuilding.
Prepare-only is not publication permission. A normal owner command to release the identified candidate
includes its required stable promotion; do not stop at a server Pre-release or ask twice for the same action.
Reuse signed prepared Manager installers after verifying their source metadata, SHA-256, updater JSON
and signature against the matching source's updater public key. Do not rerun compilation just to upload
already verified assets, or overwrite a published version with different bytes.
Server promotion uses the persistent Windows PC and the shared workflow's verified caches. Preserve
production databases; run applicable local acceptance in disposable fixtures. Record the actual passed,
failed and unrun checks. Verify both GitHub Latest and the public signed stable pointer before reporting
that users can update. Leave the durable handoff described in the runbook for the next agent.
+324 -15
View File
@@ -1,25 +1,334 @@
# CoA Server Manager changelog
Manager releases only. Updating the Manager does not update the server.
[Server changelog](https://github.com/Corfirean/coa-server-build/blob/main/CHANGELOG.md)
Published releases and previews. Pre-release entries do not change the stable update channel.
Dates follow GitHub publication (UTC); legacy records retain their archived dates. Newest releases first.
## 0.7.1 — 2026-10-22
Close the game, then open Modules → Custom Races → Repair Client Patch to update existing patches, or Install Client Patch for a new installation. Updating the Manager does not automatically replace game files or update your server.
**Portable Characters are experimental and may contain bugs.**
- Updated Custom Races client patches to the complete 1.8.3 package.
- Race patch installation now includes the hide-armor and talent-reset addons.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.7.1) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.7.0...v0.7.1)
[Guides and screenshots](https://github.com/Corfirean/coa-server-manager/blob/master/docs/README.md)
## 0.7.0 — 2026-10-21
Portable transfers into Warcraft Reborn are not supported yet.
**Portable Characters are experimental and may contain bugs.**
- Keep portable characters and their saved versions in the Manager, and use them on compatible realms.
- Make an existing character portable from a connected realm.
- Browse public realms and see their connection details.
- Link an existing realm account or create one on hosts that allow it.
- Switch between CoA, Wildcard and Warcraft Reborn realm profiles on supported server builds.
- Adjust creature health for different locations and enemy types in Content Scaling.
- Switch Custom Races patches faster, with visible feedback while changes are applied.
- Handle Windows requests for administrator permission when launching the game.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.7.0) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.22...v0.7.0)
[Player guides and screenshots](https://github.com/Corfirean/coa-server-manager/tree/d22ff09ac6fd7628aac14d90ec23f547cd9b4bf1/docs)
## 0.6.22 — 2026-10-19
- Updated Custom Races client patches to 1.8.
- Included the updated race appearance and cursor addons.
- Existing client files are preserved when installing or removing race patches.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.22) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.21...v0.6.22)
## 0.6.21 — 2026-10-16
- Server updates apply migrations once after saving a recovery point.
- Older server versions update directly without installing intermediate packages.
- Update validation recognizes an already stopped server.
- Recorded core SQL stays applied after old history tables are cleaned up.
- Switching realms no longer repeats shared account database updates.
- Update offers distinguish version support from database validation.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.21) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.19...v0.6.21)
## 0.6.19 — 2026-10-16
- Update checks no longer download a server archive to compare its launcher.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.19) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.18...v0.6.19)
## 0.6.18 — 2026-10-16
- Older servers find a tested intermediate update instead of being blocked.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.18) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.17...v0.6.18)
## 0.6.17 — 2026-10-15
- Already running MySQL is recognized before startup readiness checks.
- New servers install the current stable build directly.
- Install the matching client patches for Grummle.
- Failed repairs restore the previous server after final verification errors.
- Failed server updates show an error on the Overview screen.
- Private update checks show preparation stages and stop failure details.
- Repair recovery copies preserve all bundled database objects and accounts.
- Server checks report interrupted SQUID database imports.
- Shutdown remains available when startup templates or database records are damaged.
- Successful server repairs no longer block later server actions.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.17) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.16...v0.6.17)
## 0.6.16 — 2026-10-13
- Linux Docker servers follow Settings, sharing and backups (experimental).
- Local hosts can connect while sharing a server online.
- Custom race client installation follows the current upstream release.
- Docker friend mode requests a restart when remote client support changes.
- Module cards show installed server revisions and the client race patch version.
- Server updates accept long Windows paths inside the private test server.
- Diagnostics record blocked startup requests and maintenance owner information.
- Client race patches can be repaired and removed while preserving original files.
- Checking an already installed server update no longer reports incompatibility.
- Failed server startup shows service errors even when the launcher reported ready.
- Starting the server waits for update checks and identifies ongoing maintenance.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.16) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.15...v0.6.16)
## 0.6.15 — 2026-10-11
- Large server updates no longer repeatedly rewrite the recovery journal.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.15) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.14...v0.6.15)
## 0.6.14 — 2026-10-11
- Fixed server update checks failing with long Windows paths.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.14) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.13...v0.6.14)
## 0.6.13 — 2026-10-11
- Failed server startup after an update automatically restores the previous version.
- Windows update recovery preserves database procedures triggers and views.
- Custom races bot setting synchronizes Playerbots excluded race lists.
- Blocked updates identify database objects that prevent automatic recovery.
- Updates can follow signed channels that select a complete server release.
- Server updates check startup privately before allowing players to connect.
- Updates repair missing database defaults from the verified release schema.
- Server updates no longer replay module installation scripts on existing databases.
- Windows server updates are tested on a private copy before changing your installation.
- Server updates can require a tested combination of platform and installed version.
- Failed database checks keep a complete report after rollback.
- Players can install custom race patches when connecting to a friend.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.13) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.12...v0.6.13)
## 0.6.12 — 2026-10-10
- Background world servers no longer stop because the local console is closed.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.12) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.11...v0.6.12)
## 0.6.11 — 2026-10-10
- Server and Bots settings open without a blank window.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.11) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.10...v0.6.11)
## 0.6.10 — 2026-10-10
- Custom race client patches update to the latest races and appearances.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.10) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.9...v0.6.10)
## 0.6.9 — 2026-10-10
- Custom races can be enabled and their client patches installed from Modules.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.9) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.8...v0.6.9)
## 0.6.8 — 2026-10-10
- Linux: install from the install screen, play with Wine or Proton (experimental).
- A failing database update is no longer reported as "database not running.
- Unsaved settings are protected and crashed servers can restart automatically.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.8) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.7...v0.6.8)
## 0.6.7 — 2026-10-08
- Choose where backups are stored, for example on another drive.
- Bot settings with a fixed list of values show a drop-down.
- The bot dashboard loads completely and can be opened in the browser.
- Accounts of the SQUID bots are hidden from the Accounts list.
- A Dashboard tab shows what the SQUID bots are doing.
- A stopped server stays stopped after an update, unless you ask it to start.
- Checking for updates no longer fails with "Something went wrong" after a crash.
- Server updates no longer stop with "has a different checksum" on the Wildcard repair.
- A world server started from the coa-bots folder is recognised.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.7) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.6...v0.6.7)
## 0.6.6 — 2026-10-07
- Install a server from a package on Linux, running in Docker (experimental).
- The Auction House Bot card no longer asks for a character GUID.
- Choose and remember the local network address friends use to connect.
- The NPC Enchanter card explains how to place and remove the NPC.
- Diagnostic files with database checks stay readable after secrets are removed.
- One button collects all diagnostics, including crashes and Windows crash records.
- Problem reports go to the right project.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.6) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.5...v0.6.6)
## 0.6.5 — 2026-10-06
- Imported SquidBots servers warn before replacing their original binary.
- SquidBots settings follow the options shipped by the module.
- SquidBots shows its bundled release and includes it in problem reports.
- Server updates accept verified published migration variants without rerunning them.
- Server updates replace older launchers adapted by the Manager for Wildcard.
- Older server packages cannot replace a newer installation.
- Checking again no longer offers an already installed launcher update.
- Large SQUID database imports have more time to finish on first startup.
- Update history conflicts are detected before server files change.
- Pending database updates remain visible when the server version matches.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.5) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.4...v0.6.5)
## 0.6.4 — 2026-10-05
- Start your installed server or client without downloading an available update.
- Finish a server update that was blocked by a startup failure.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.4) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.3...v0.6.4)
## 0.6.3 — 2026-10-04
- Configure partial dungeon parties and open-world life steal in module settings.
- Download and update the game client without installing a server.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.3) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.2...v0.6.3)
## 0.6.2 — 2026-10-04
- Choose which item types the auction bot sells.
- Module settings now offer translated controls and an enemy damage slider.
- Failed server updates restore databases and files together.
- Configure SQUID Playerbots and prevent conflicting bot systems.
- Server packages must include a complete database schema check.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.2) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.1...v0.6.2)
## 0.6.1 — 2026-10-04
- PLAY launches only the client with the selected realmlist; starting the local server remains a separate action.
- Server multipliers accept decimals typed with a dot or comma. Talent and rested experience descriptions now explain their actual behavior.
- Added database and file checks, and Repair with backups and signed package verification. Unresolved schema damage is reported instead of guessed repairs.
- Fixed character database rebuilds retaining migration history that could skip required SQL. Applied migration hash changes are now rejected.
- Added optional simultaneous CoA and Wildcard worlds on Windows, with separate processes, ports and status indicators.
- Server multipliers now accept decimals typed with a dot or comma.
- Play launches only the game client without starting your local server.
- Check database structure and repair server files with a safety backup.
- Start CoA and Wildcard together from Settings.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.1) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.6.0...v0.6.1)
## 0.6.0 — 2026-10-03
- Added Simplified Chinese for the interface and all server and bot settings.
- The closed game client now follows the selected CoA or Wildcard realm. Play checks the saved realm before launching; existing settings are preserved and backed up.
- Fixed GitHub reports opening Windows Explorer instead of the default browser. Long reports use the clipboard fallback.
- Fixed configuration snapshots created in the same millisecond overwriting each other during restore.
- Added experimental Linux startup: persistent state uses the XDG data folder, Tailscale detection supports Linux, and a Wayland rendering workaround respects existing user settings.
- Added experimental start, stop and status controls for manually prepared Docker servers, with private database networking and a remote console bound to localhost.
- Linux extraction restores executable permissions for ELF binaries and scripts, and paths retain case-sensitive filesystem semantics.
- The game client follows the realm selected in the Manager.
- Configuration snapshots no longer overwrite each other.
- Experimental Linux startup and Docker server controls.
- Problem reports open in the default browser.
- Simplified Chinese is available in the Manager.
### Availability
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.0) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.5.0...v0.6.0)
This release ships the Windows installer and signed updater assets. Linux support is a development preview, not a one-click installation flow. Docker installation/database setup, SQL operations and backups, updates, Wildcard realm profiles, and Wine/Proton client launching remain unfinished. Servers must be prepared manually and use Linux binaries built with the packaged configuration layout.
## 0.5.0 — 2026-10-03
Server shutdown and starting-zone scaling issues require separate server releases and are not fixed by updating the Manager alone.
- Imported servers get the module config files they lack.
- Modules marked Soon stay greyed out.
- The Bots page is hidden when the module is not on the server.
- Updating the game client shows a progress bar on its button.
- Switch between CoA and Wildcard while keeping separate characters and module settings.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.5.0) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.4.0...v0.5.0)
## 0.4.0 — 2026-10-02
- Added module cards with switches and settings.
- Added searchable server settings and appearance and collection controls.
- Added account deletion with confirmation.
- Fixed update checks offering an already installed version.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.4.0) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.3.1...v0.4.0)
## 0.3.1 — 2026-10-01
- Added named realmlists and a switcher beside Play.
- Added account management and a prefilled GitHub problem-report form.
- Server updates continue when leaving the settings page.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.3.1) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.3.0...v0.3.1)
## 0.3.0 — 2026-10-01
- Download and update the game client through the Manager while preserving custom files.
- Server updates are checked automatically and offered on Overview.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.3.0) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.2.0...v0.3.0)
## 0.2.0 — 2026-10-01
- Added companion creation progress and controls to stop creation, take bots offline or remove them.
- Choose how many bots log in when the server starts.
- Added a control to log out a chosen number of bots and reduce server load.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.2.0) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.1.6...v0.2.0)
## 0.1.6 — 2026-09-30
- Overview shows world speed.
- Improved companion setup, including requests made while the server is stopped.
- Expanded translated settings and connection guides.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.6) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.1.4...v0.1.6)
## 0.1.4 — 2026-09-30
- Sharing settings are reapplied before server startup.
- Added a restart shortcut after enabling play with friends.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.4) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.1.3...v0.1.4)
## 0.1.3 — 2026-09-30
- The interface scales with the window.
- Each friend connection mode shows the realmlist line to use.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.3) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.1.2...v0.1.3)
## 0.1.2 — 2026-09-30
- Manager updates download in the background and install when the app closes.
- Missing module configuration files are created from the supplied defaults.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.2) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.1.1...v0.1.2)
## 0.1.1 — 2026-09-30
- Added translations throughout setup, server updates, backups, diagnostics and settings.
- Added clearer errors when server packages are unavailable or a server cannot be reached.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.1) · [Source changes](https://github.com/Corfirean/coa-server-manager/compare/v0.1.0...v0.1.1)
## 0.1.0 — 2026-09-30
- Initial Windows release: install or import a server and manage startup from one app.
- Configure server and bot settings, manage accounts, create backups and restore saved data.
- Launch the game client, set up sharing with friends and collect diagnostics.
[Release](https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.0)
Generated
+3 -3
View File
@@ -491,7 +491,7 @@ dependencies = [
[[package]]
name = "coa-core"
version = "0.6.15"
version = "0.6.21"
dependencies = [
"base64 0.22.1",
"chrono",
@@ -520,7 +520,7 @@ dependencies = [
[[package]]
name = "coa-release"
version = "0.6.15"
version = "0.6.21"
dependencies = [
"chrono",
"coa-core",
@@ -529,7 +529,7 @@ dependencies = [
[[package]]
name = "coa-server-manager"
version = "0.6.15"
version = "0.6.21"
dependencies = [
"coa-core",
"serde",
+1 -1
View File
@@ -3,7 +3,7 @@ resolver = "2"
members = ["crates/coa-core", "crates/coa-release", "src-tauri"]
[workspace.package]
version = "0.6.15"
version = "0.6.21"
edition = "2021"
license = "AGPL-3.0-only"
+34 -16
View File
@@ -1,12 +1,16 @@
# CoA Server Manager
**Portable characters are experimental.** This feature is in active development and may contain bugs.
A Windows app that installs and runs your own **Conquest of Azeroth** server (AzerothCore) with a few clicks:
no console, no config files, no database tools. Tauri 2 + Rust + React.
![Overview](docs/screenshots/play.png)
![Your server at a glance](docs/screenshots/overview.png)
## What it does
* **Portable characters (Experimental).** Keep saved characters in the Manager, review their saved versions and make an existing character portable from a realm you can access. See the [short guide](docs/PORTABLE-CHARACTERS.md).
* **Find a realm.** Browse public realms and compare their game modes, level caps, rates, modules and population.
* **One-click install.** Downloads a signed, verified server package (resumable, checked file by file), sets up the
database and creates your first account. Or import a server you already have.
* **Start / Stop / PLAY.** One button for the whole server (database, auth, world) with live status, uptime and
@@ -18,14 +22,14 @@ no console, no config files, no database tools. Tauri 2 + Rust + React.
into **UPDATE**. Updates are authenticated (Ed25519), back up first, keep files you changed and can be rolled back.
* **Backups.** Characters, accounts and settings on demand, automatically before every update; verify and restore
from the app.
* **CoA / Wildcard realms.** Choose the active realm on Overview. A server build with Wildcard support is required;
* **CoA / Wildcard / Warcraft Reborn realms.** Choose the active realm on Overview. A server build supporting the selected realm is required;
the first switch copies world data and creates an empty character database. Accounts are shared, characters and
module settings are separate, and only one world runs. Switching a running server saves characters and restarts it.
The closed client's saved realm follows this selection and is checked again before PLAY. If the client is open,
the change is applied on the next launch through the Manager.
CoA companions are unavailable on Wildcard; other optional modules are experimental. Backups include both realms,
and database updates apply to both.
See [realm profiles and validation](docs/WILDCARD.md).
Bots are unavailable on Wildcard and Warcraft Reborn; other optional modules are experimental. Backups and
database updates include every created realm. Simultaneous startup supports CoA + Wildcard only.
See [Wildcard guide](docs/WILDCARD.md) and [Warcraft Reborn](docs/WARCRAFT-REBORN.md).
* **Companions (bots).** Fill the world with level-appropriate bots, see progress, stop spawning, take them offline,
despawn N to relieve the server, and choose how many log in at server start.
* **Realmlist switcher.** Keep named realmlists (for example *Solo* and *PTR*) and switch the game client between them from a
@@ -39,17 +43,22 @@ no console, no config files, no database tools. Tauri 2 + Rust + React.
supports LAN, direct and private-network (Tailscale) play, with step-by-step guides and a package to send to friends.
* **Updates to the Manager itself** download in the background and install when you close it. Your servers are never
stopped or changed by a Manager update.
* **Languages:** English, Russian, German, French, Spanish.
* **Languages:** English, Russian, German, French, Spanish, Simplified Chinese.
| Install | Ready |
|---------|-------|
| ![Installing](docs/screenshots/install.png) | ![Ready](docs/screenshots/ready.png) |
## See the Manager
| Overview | Server updates |
|----------|----------------|
| ![Overview](docs/screenshots/overview.png) | ![Updates](docs/screenshots/updates.png) |
| Portable characters (Experimental) | Public realms |
|---|---|
| ![Portable characters](docs/screenshots/portable.png) | ![Public realms](docs/screenshots/realms.png) |
![Backups](docs/screenshots/backups.png)
| Companions | Backups |
|---|---|
| ![Companions](docs/screenshots/bots.png) | ![Backups](docs/screenshots/backups.png) |
The full gallery also covers setup, modules, game clients, playing with friends and making an existing character portable.
[Browse the screenshot gallery](docs/SCREENSHOTS.md) · [Player and server owner guides](docs/README.md).
## Install
@@ -57,11 +66,20 @@ Download the installer from the [Releases](https://github.com/Corfirean/coa-serv
and run it. You need about 10 GB of free disk space for the server. The game client (about 43 GB) is separate: point the
Manager at the one you have, or let it download one for you.
## Changelog
[Manager release history](CHANGELOG.md) · [Server release history](https://github.com/Corfirean/coa-server-build/blob/main/CHANGELOG.md).
Manager and server updates have separate version histories.
## Status
Phases 0–10 of the plan in [docs/ARCHITECTURE.md](docs/ARCHITECTURE.md) are done: foundation, import, start/stop,
config engine, backups, clean install from signed packages, updates, companions, client launch, friends, polish.
Remaining work (translation review by native speakers, optional extras) is listed in section 9 of that document. Translation notes: [docs/TRANSLATIONS.md](docs/TRANSLATIONS.md).
**In active development.** Bugs may still occur. If you run into a problem, please [open an issue](https://github.com/Corfirean/coa-server-manager/issues/new/choose).
### Known Limitations
* **Native client login prefill**: Ascension's 3.3.5 client does not accept command-line credentials; users type their password once or link an existing account.
* **Non-Windows SecretStore**: Windows uses DPAPI (`CryptProtectData`) with domain separation entropy. On Linux/macOS, keys fall back to strict 0600 file permissions.
* **Direct to Relay migration**: Established direct TCP sessions cannot migrate mid-stream to Relay; network breaks trigger automatic fallback on the next JOIN.
* **WorkerThreads**: `CharacterDatabase.WorkerThreads` must be set to `1` (the default) for safe remote portable character imports.
## Develop
```
@@ -0,0 +1,7 @@
---
area: manager
type: added
audience: admins
title: Linux Docker servers follow Settings, sharing and backups (experimental)
---
Experimental. Settings the Docker backend fixes are hidden, friends sharing and logs work, the game data folder can be changed, and the realm no longer shows offline after a start.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Local hosts can connect while sharing a server online
---
The Manager no longer replaces the realm's local address with its public address when Friends mode is enabled.
@@ -0,0 +1,7 @@
---
area: client
type: changed
audience: admins
title: Custom race client installation follows the current upstream release
---
The installer verifies the current race archives and includes the two additional MPQs. Legacy patch hashes remain available so upgrading never mistakes the old custom executable for an original launcher file.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Docker friend mode requests a restart when remote client support changes
---
Enabling remote CoA clients now requests a world restart even when the game ports were already open. Repeating the same unchanged setting does not request another restart.
@@ -0,0 +1,7 @@
---
area: manager
type: added
audience: admins
title: Module cards show installed server revisions and the client race patch version
---
Module revisions come from the currently installed release metadata. The race card shows the client patch version separately; missing version metadata is reported honestly instead of showing a download version as installed.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Server updates accept long Windows paths inside the private test server
---
Long SQL file paths no longer fail the folder safety check during update rehearsals. Files outside the server folder remain blocked.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Diagnostics record blocked startup requests and maintenance owner information
---
A start refused before the launcher runs is now logged with its reason. Exported diagnostics include the last recorded maintenance owner, while clearly distinguishing old metadata from a currently held operating-system lock.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Checking an already installed server update no longer reports incompatibility
---
The installed release can be checked again even when its compatibility list only names earlier releases. Platform checks and protection against downgrades still apply.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Failed server startup shows service errors even when the launcher reported ready
---
Update rehearsal and startup failures include recent world, authentication and database error lines. Missing tables and late startup failures are visible directly in the error details, with secrets redacted.
+7
View File
@@ -0,0 +1,7 @@
---
area: bots
type: fixed
audience: admins
title: Disabling custom races for SQUID bots now excludes all nonstandard races
---
The standard-races preset keeps only the ten original playable races. High elves, worgen and other custom race IDs are excluded, and the displayed setting reads complete race IDs rather than matching number substrings.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Starting the server waits for update checks and identifies ongoing maintenance
---
Downloading an update manifest no longer blocks startup. A start waits up to 30 seconds for a check to finish; other active maintenance still blocks it safely and shows its operation, process and start time.
@@ -0,0 +1,7 @@
---
area: bots
type: changed
audience: admins
title: The experimental SQUID race preset follows the current race release
---
Custom-race bots now use the upstream supported-race preset, excluding Test Race, Skyborne and Darkfallen. The standard-races preset continues to exclude all custom races.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Already running MySQL is recognized before startup readiness checks
---
A verified MySQL listener replaces a stale or supervisor-only launcher record before startup. Existing databases are reused instead of waiting for a port owned by a different process from the same executable.
@@ -0,0 +1,7 @@
---
area: manager
type: changed
audience: admins
title: New servers install the current stable build directly
---
A clean installation downloads the complete package of the selected stable release, including its prepared database. It no longer installs an old base and asks for a separate update. Missing or mismatched full packages stop installation before changing the destination.
@@ -0,0 +1,7 @@
---
area: manager
type: changed
audience: players
title: Install the matching client patches for Grummle
---
Custom race installation includes the new Grummle appearance and voices while retaining the existing race assets. Older installations can still be safely switched off or removed.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Failed repairs restore the previous server after final verification errors
---
Errors after file replacement now undo the repair using its verified recovery point, including final file or database check errors. If recovery fails, the journal retains the cause and further server actions remain blocked.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Failed server updates show an error on the Overview screen
---
The error stays visible after the progress bar disappears, with technical details and a copy button. Recovery-state errors are shown there too.
@@ -0,0 +1,7 @@
---
area: manager
type: changed
audience: admins
title: Private update checks show preparation stages and stop failure details
---
Private-copy preparation reports checked and copied file counts, database tests and startup stages. Failed shutdown reports the server folder, remaining processes and launcher reason. Port conflicts identify the owning process and executable.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Repair recovery copies preserve all bundled database objects and accounts
---
Windows repairs now save a verified copy of stopped MySQL before changing server files or SQL. Recovery preserves procedures, triggers, views and database accounts as well as player rows.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Server checks report interrupted SQUID database imports
---
Check and Repair now report SQUID pending imports and missing bot character tables separately from package migrations. Interrupted imports retain their records and require verified database recovery.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Shutdown remains available when startup templates or database records are damaged
---
Stop uses existing shutdown settings instead of rendering startup templates. Missing or stale MySQL records are recovered only from the exact installed executable and its owned port; foreign processes are never adopted.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Successful server repairs no longer block later server actions
---
Repair verifies restored files before completing its recovery record. Older completed repair records are accepted only when every repaired file still matches its expected content.
@@ -0,0 +1,7 @@
---
area: manager
type: changed
audience: admins
title: Server updates apply migrations once after saving a recovery point
---
Updates no longer build and start a second full server copy. The Manager checks both realms, saves recovery, applies pending SQL once and verifies startup. Failed SQL or startup restores the recovery point. Full-copy upgrade testing remains a release task.
@@ -0,0 +1,7 @@
---
area: manager
type: changed
audience: admins
title: Older server versions update directly without installing intermediate packages
---
The Manager checks database history and applies missing SQL from the latest signed package. A version missing from the release test list no longer blocks the update. Unknown versions, interrupted migrations and downgrades remain blocked.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Update validation recognizes an already stopped server
---
Validation uses the same complete-server stop check as recovery instead of issuing redundant unconditional shutdown commands.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Older servers find a tested intermediate update instead of being blocked
---
When the latest release cannot update an older server directly, Manager offers a signed intermediate release from its compatibility matrix and explains how to continue to the latest version.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Update checks no longer download a server archive to compare its launcher
---
Manager recognizes its supported launcher edits against the signed file hash locally. Overlapping checks wait briefly for one another, and the private update copy checks combined program and database space before copying.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Recorded core SQL stays applied after old history tables are cleaned up
---
Existing core migration records are preserved in Manager history. Unknown legacy checksums remain explicitly unknown, and interrupted or changed migrations are still rejected.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Switching realms no longer repeats shared account database updates
---
Solo and Wildcard use one migration history for their shared account database. Previously applied records from either realm are recognized without running their SQL again.
@@ -0,0 +1,7 @@
---
area: manager
type: fixed
audience: admins
title: Update offers distinguish version support from database validation
---
An available version sequence no longer claims that every database state has been tested. Database compatibility is checked during the update.
@@ -0,0 +1,7 @@
---
area: manager
type: added
audience: admins
title: Module settings for Content Scaling include creature health
---
The Content Scaling page now has five health multipliers: for the open world, dungeons and raids, and for ordinary mobs and bosses. They multiply the health the scaling already gives creatures (1 keeps the normal balance, lower is easier, 0.1 to 10) and need a server build that includes the matching Content Scaling settings.
@@ -0,0 +1,7 @@
---
area: manager
type: added
audience: admins
title: Choose Parallel or Global level scaling in module settings
---
Content Scaling progression can now be set to Parallel (Classic over 1 to the level cap, TBC and WotLK over 45 to the cap) or Global (all expansions over 1 to the cap) without editing the config file by hand.
+1 -2
View File
@@ -1,7 +1,6 @@
# changelog.d
One small Markdown file per change that a player or a server owner would notice. They are collected from every CoA
repository when a release is made. The rules and the file format are in `AGENTS.md` (section "Changelog") and, in full,
One small Markdown file per change that a player or a server owner would notice. Server and Manager releases collect only their own shipped changes from immutable sources; their public histories are separate. The rules and the file format are in `AGENTS.md` (section "Changelog") and, in full,
in https://github.com/Corfirean/coa-server-build/blob/main/changelog/GUIDE.md
This README is ignored by the collector.
@@ -0,0 +1,200 @@
[
{
"tag": "v0.6.22",
"date": "2026-10-19T07:49:11Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.22",
"body": "## Manager\r\n- Updated Custom Races client patches to 1.8, including Drakkin, Drakonid and updated racial abilities\r\n- Added updated race appearances and cursor addons\r\n- Existing client files are preserved when installing or removing race patches\r\n\r\nPrepared from commit 65df6604846d5492e40a6b08a797199478666e24. Installer and updater signature were verified before publication.\r\n"
},
{
"tag": "v0.6.21",
"date": "2026-10-16T12:11:35Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.21",
"body": "## Manager\n- Older server versions update directly to the latest signed package without installing intermediate releases.\n- The published source-version list records release test coverage; a missing entry no longer blocks an older installation by itself.\n- Apply only missing database migrations, preserving recorded AzerothCore history and sharing account migration history between Solo and Wildcard.\n- Save one complete recovery point, apply SQL once and check the database schema, startup, account and character identities. Restore files and databases on failure.\n- Keep unknown imported versions, interrupted migrations, changed SQL checksums, downgrades and incompatible packages blocked.\n- Remove the nested per-update server rehearsal and recursive intermediate-package lookup.\n\nInstalling this Manager does not update your server automatically. Unknown third-party repacks still require a verified starting state; this release does not guess which unrecorded SQL was applied manually.\r\n\nValidation: 11 package-based scenarios passed against Server 0.261015.58, including the earliest base and a repeated current-version check. A separate direct 800006 -> 58 upgrade on an isolated copy preserved all 133 account and 3123 character identities. No client login or gameplay acceptance was performed.\n\r\n"
},
{
"tag": "v0.6.19",
"date": "2026-10-16T02:37:16Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.19",
"body": "## Manager\n- Update checks no longer download a full server package just to verify the launcher\n- Concurrent update checks wait briefly for the current check instead of immediately reporting another operation\n- Update rehearsals check the combined space required for program files and databases before creating the private copy\n\nIncludes the signed intermediate upgrade path from v0.6.18 for older server installations.\n\nThis release updates the Manager only. Existing server files and databases are not changed by installing it.\r\n"
},
{
"tag": "v0.6.18",
"date": "2026-10-16T00:33:21Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.18",
"body": "## CoA Server Manager 0.6.18\n\n- Fixed update checks blocking older server versions when a tested intermediate release is available\n- Manager now offers the required intermediate update automatically and explains how to continue to the latest stable version\n- Intermediate packages must match the signatures and manifest hashes recorded in the target release's compatibility matrix\n- Unsupported source errors now identify the installed version, target version and directly supported versions\n\nFor servers on **0.261011.800006**, Manager offers **0.261013.54** first. After it completes, check for updates again to install **0.261015.58**. Clear any manually entered Advanced package source to use automatic selection.\n\nThe server package and stable channel have not changed. Existing data follows the normal backup and verified update process.\n"
},
{
"tag": "v0.6.17",
"date": "2026-10-15T16:51:52Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.17",
"body": "## CoA Server Manager 0.6.17 / Server 0.261015.58\n\n### Manager\n\n- New servers now install the current stable build directly, without installing 2.0 first\n- Fixed completed repairs leaving the server stuck in a maintenance state\n- Servers can now stop even when startup templates or MySQL process records are damaged\n- Repair now preserves database accounts, views, procedures and triggers, with improved rollback after failures\n- Interrupted SQUID database imports are now reported by Check and Repair\n- Update errors remain visible on Overview, with technical details and a copy button\n- Improved update preparation progress and diagnostics for shutdown failures and port conflicts\n- Updated Custom Races client patches to 1.7 while preserving existing race assets\n\n### Server\n\n- Integrated 34 CoA upstream commits\n- Updated Custom Races to 1.7, adding playable Grummle and their racial abilities\n- Updated SQUID Playerbots to 2.3.1\n- Added complete installation packages for the current server release\n- Improved shutdown recovery and startup configuration errors\n\n### Content scaling\n\n- Creature health settings are now respected while Content Scaling is enabled\n"
},
{
"tag": "v0.6.16",
"date": "2026-10-13T12:39:15Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.16",
"body": "**Manager** (for server owners)\n- Added: Linux Docker servers follow Settings, sharing and backups (experimental). Experimental. Settings the Docker backend fixes are hidden, friends sharing and logs work, the game data folder can be changed, and the realm no longer shows offline after a start.\n- Added: Updates can follow signed channels that select a complete server release. The Manager checks the channel signature and downloads every package part from the same release. Existing installations continue using the previous download source until the new channel is published.\n- Added: Windows server updates are tested on a private copy before changing your installation. The Manager tests migrations and server startup using your databases and configurations. A failed rehearsal leaves the installed server unchanged and preserves a diagnostic report.\n- Added: Server updates can require a tested combination of platform and installed version. Signed release metadata identifies core and module revisions, supported starting versions and database contracts. Incompatible packages are rejected before shutdown; changed schema contracts and incompatible databases block installation before migration.\n- Added: Module cards show installed server revisions and the client race patch version. Module revisions come from the currently installed release metadata. The race card shows the client patch version separately; missing version metadata is reported honestly instead of showing a download version as installed.\n- Changed: Failed server startup after an update automatically restores the previous version. The Manager restores server files and databases, then checks the recovered server. If recovery cannot be confirmed, it preserves the recovery point and blocks further startup until the problem is resolved.\n\n----- next message -----\n\n**Manager** (for server owners) (cont.)\n- Fixed: Windows update recovery preserves database procedures triggers and views. A verified copy of stopped MySQL restores all databases and accounts together. Interrupted restoration can be retried, and the replaced database directory is retained.\n- Fixed: Custom races bot setting synchronizes Playerbots excluded race lists. The \"Bots can be custom races\" setting now configures AiPlayerbot.ExcludedBotRaces in playerbots.conf. Off excludes all added races (recommended), while on allows stable added races (experimental).\n- Fixed: Blocked updates identify database objects that prevent automatic recovery. Diagnostics include the complete list of unsupported database objects across the recovery point, helping resolve update failures without removing user data.\n- Fixed: Server updates check startup privately before allowing players to connect. Startup checks use temporary local ports. Original settings are restored before opening the server, and interrupted checks retain a recovery journal.\n- Fixed: Local hosts can connect while sharing a server online. The Manager no longer replaces the realm's local address with its public address when Friends mode is enabled.\n- Fixed: Updates repair missing database defaults from the verified release schema. Only missing literal defaults are restored when column properties match. Existing rows, custom defaults, types and nullability are preserved; unresolved incompatibilities still block the update.\n- Fixed: Server updates no longer replay module installation scripts on existing databases. Update packages contain incremental SQL only. Module table initialization and seed scripts remain available when preparing a clean installation, preventing existing content from being reset during updates.\n\n----- next message -----\n\n**Manager** (for server owners) (cont.)\n- Fixed: Failed database checks keep a complete report after rollback. Schema errors show readable default values and save every detected difference for troubleshooting, including after the server is restored.\n- Fixed: Large server updates no longer repeatedly rewrite the recovery journal. Updates containing rebuilt navigation data now prepare all recovery copies before replacing files. Interrupted updates remain recoverable without rewriting the full journal for every file.\n- Fixed: Normalize rehearsal server roots to prevent Windows path prefix errors. Server update rehearsals now canonicalize the installation root before collecting files, resolving a \"prefix not found\" error during rehearsal preparation on Windows.\n- Fixed: SQUID imports character tables first and reports slow database imports. SQUID prepares character tables before its large travel data. Startup shows each base import and identifies a failed import without starting the world server or replaying an incomplete migration.\n- Fixed: Docker friend mode requests a restart when remote client support changes. Enabling remote CoA clients now requests a world restart even when the game ports were already open. Repeating the same unchanged setting does not request another Line truncated
},
{
"tag": "v0.6.15",
"date": "2026-10-11T23:10:59Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.15",
"body": "## Fixes\n\n- Large server updates, including rebuilt maps, vmaps and mmaps, no longer rewrite the complete recovery journal for every file.\n- All recovery copies are prepared before replacing any server file. Interrupted updates still restore the previous files and remove newly created files only when their content matches the update.\n\nValidated by Windows and Linux CI, the full Manager test suite, and a regression covering power loss during file replacement and incomplete recovery-copy preparation.\n\nUpdate the Manager before installing the navigation refresh. Portable Characters and Portable Session Bridge are not included."
},
{
"tag": "v0.6.14",
"date": "2026-10-11T21:14:34Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.14",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.6.13",
"date": "2026-10-11T14:03:44Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.13",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.6.12",
"date": "2026-10-10T03:10:14Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.12",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.6.11",
"date": "2026-10-10T03:05:20Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.11",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.6.10",
"date": "2026-10-10T02:07:18Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.10",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.6.9",
"date": "2026-10-10T01:50:51Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.9",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.6.8",
"date": "2026-10-10T01:05:56Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.8",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.6.7",
"date": "2026-10-08T04:59:31Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.7",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.6.6",
"date": "2026-10-07T05:47:46Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.6",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.6.5",
"date": "2026-10-06T06:17:21Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.5",
"body": "# CoA Server Manager 0.6.5 / Server 0.261006.1\n\nUpdate CoA Server Manager to 0.6.5 first, then update the server. This server package requires Manager 0.6.5.\n\n## Server updates and recovery\n\n- Recognize known, signed historical migration checksums without rewriting migration history or replaying applied SQL.\n- Check migration history and SQL files before replacing server files. Interrupted or failed migrations still require recovery.\n- Preserve damaged update records and protect recovery points belonging to unfinished updates. Check saved copies before restoring them.\n- Refresh saved update status after update and recovery attempts, including failures.\n- Recognize launcher adjustments made by the Manager, including older Wildcard launchers. Custom edits still require a choice.\n- Stop offering an already installed launcher update after checking again.\n- Reject older server packages. Use the update's recovery point to undo an update.\n- Restore missing Wildcard card and specialization-cache tables through a new corrective migration, preserving existing tables and data.\n- Fix launcher integration imports when using the server's isolated Python runtime.\n\n## SQUID Playerbots\n\n- Include SQUID Playerbots v1.8.1 together with the matching core, launcher and Manager integration.\n- Provide translated bot settings, release information and controls for the active bot system. Prevent conflicting bot systems from starting together.\n- Include the SQUID database in recovery points when installed.\n- Keep servers with SQUID disabled independent of its database startup hooks.\n- Detect unfinished initial database imports instead of treating partially imported tables as complete.\n- Allow more time for large SQL imports and initial bot creation. The Manager startup deadline is up to 45 minutes when SQUID is enabled; ordinary startup deadlines are unchanged.\n- Preserve historical SQL files and migration history. Only new migrations are applied.\n- Require an explicit keep-or-replace decision when updating imported server binaries protected by their original launcher or updater.\n\n## Local candidate validation\n\nThe signed final package passed an update and startup of both CoA and Wildcard on a marked disposable installation. Selected account, character and inventory fields were preserved, historical migrations were skipped, and a repeated update check was clean. Fresh SQUID database import and repeated startup were checked using the same native binaries and SQL as the final package.\n"
},
{
"tag": "v0.6.4",
"date": "2026-10-05T09:22:40Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.4",
"body": "Fixes server updates that failed to start with the repack's bundled Python runtime.\n\nIf an update is waiting for recovery, open Settings → Server updates and choose **Retry server startup**. Successful validation finishes the already applied update without replaying database migrations. Rollback remains available if validation fails.\n\n**Start without updating** and **Play without updating** let you use the installed server or client when a newer version is available. Interrupted server updates still require recovery.\n\n[Shared changelog](https://github.com/Corfirean/coa-server-build/blob/main/CHANGELOG.md)\n"
},
{
"tag": "v0.6.3",
"date": "2026-10-04T22:41:10Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.3",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.6.2",
"date": "2026-10-04T21:54:32Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.2",
"body": "## Manager 0.6.2\n\n- Server Update and Repair now back up databases before changes and recover databases, files and version metadata together after a failed update.\n- Database checks detect missing Wildcard tables even when migration history says the SQL ran. New server packages include the expected full database schema.\n- Interrupted or failed SQL is not blindly replayed; unresolved recovery is reported before another update or server start.\n- Checks cover race/class starting data and existing character visibility.\n- Adds the bot and module settings included in the reviewed dependency, including SQUID settings and translated controls.\n- Includes the latest Docker database-access changes.\n\nInstall/update Manager first, then update the server package. Run Settings → Check files and database if an existing installation reports database problems. Already lost database rows require a backup; restoring table structure cannot recreate them.\n\nValidation: 233 core tests passed locally after merging current master; workspace check and frontend build passed. Release package verification used disposable MySQL and a signed server package.\r\n\n\nShared release changelog: https://github.com/Corfirean/coa-server-build/blob/main/CHANGELOG.md\n\r\n"
},
{
"tag": "v0.6.1",
"date": "2026-10-04T10:58:56Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.1",
"body": "## Changes\n\n- PLAY launches only the game client with the selected realmlist. Use START SERVER separately for your local server.\n- Decimal server multipliers can be typed with either a dot or a comma.\n- Talent settings explain that the standard AzerothCore multiplier does not affect CoA talent points; rested experience settings describe accumulation correctly.\n- Added file/database checks and Repair with backups, verified official packages and reports of unresolved schema damage.\n- Fixed stale character migration history after rebuilding a character database and reject changed hashes of already applied migrations.\n- Added optional simultaneous CoA and Wildcard worlds on Windows, with separate processes, ports, logs and status indicators.\n\n## Notes\n\nRepair requires a stopped server and matching currently published packages. It preserves player data and settings; unsupported schema repairs are reported for corrective migrations.\n\nSimultaneous worlds require a compatible Wildcard server build. Existing Linux/Docker support remains experimental; simultaneous worlds are Windows-only.\n\nDownload the Windows installer below, or let an existing Manager installation update automatically. Server packages are updated separately.\r\n"
},
{
"tag": "v0.6.0",
"date": "2026-10-03T18:06:31Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.0",
"body": "# CoA Server Manager changelog\r\n\r\n## 0.6.0 — 2026-10-03\r\n\r\n- Added Simplified Chinese for the interface and all server and bot settings.\r\n- The closed game client now follows the selected CoA or Wildcard realm. Play checks the saved realm before launching; existing settings are preserved and backed up.\r\n- Fixed GitHub reports opening Windows Explorer instead of the default browser. Long reports use the clipboard fallback.\r\n- Fixed configuration snapshots created in the same millisecond overwriting each other during restore.\r\n- Added experimental Linux startup: persistent state uses the XDG data folder, Tailscale detection supports Linux, and a Wayland rendering workaround respects existing user settings.\r\n- Added experimental start, stop and status controls for manually prepared Docker servers, with private database networking and a remote console bound to localhost.\r\n- Linux extraction restores executable permissions for ELF binaries and scripts, and paths retain case-sensitive filesystem semantics.\r\n\r\n### Availability\r\n\r\nThis release ships the Windows installer and signed updater assets. Linux support is a development preview, not a one-click installation flow. Docker installation/database setup, SQL operations and backups, updates, Wildcard realm profiles, and Wine/Proton client launching remain unfinished. Servers must be prepared manually and use Linux binaries built with the packaged configuration layout.\r\n\r\nServer shutdown and starting-zone scaling issues require separate server releases and are not fixed by updating the Manager alone.\r\n"
},
{
"tag": "v0.5.0",
"date": "2026-10-03T06:26:08Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.5.0",
"body": "## CoA and Wildcard realm profiles\n\nChoose CoA or Wildcard on Overview. The first Wildcard switch creates a separate world and empty character database; accounts are shared. Only one world runs at a time. Switching a running server saves characters and restarts it, and returning to CoA restores its module settings.\n\n- Requires a server build and database migrations with Wildcard support. Existing destination databases or realm ID 2 are preserved and block automatic creation.\n- CoA companion bots are unavailable on Wildcard. Content Scaling and other optional modules remain experimental there.\n- Backups cover both realms; restores identify their realm. Server migrations update both realms and shared auth once.\n- Configuration switches support recovery after interruption.\n\nValidation: 178 Rust tests, frontend build and translation checks, native isolated-fixture creation/start/switch, full backup, restore and migration checks. Final real-clock Wildcard gameplay verification: four scenarios passed with zero failures and zero isolated reruns (first login, level rolls, skill-card rolls, stats). Module gameplay compatibility beyond those scenarios remains unverified.\n\nDownload the Windows installer below. Existing Manager installations receive the signed update automatically. [Details and demo](https://github.com/Corfirean/coa-server-manager/blob/v0.5.0/docs/WILDCARD.md).\r\n\r\n"
},
{
"tag": "v0.4.0",
"date": "2026-10-02T16:51:50Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.4.0",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.3.1",
"date": "2026-10-01T11:44:28Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.3.1",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.3.0",
"date": "2026-10-01T03:13:59Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.3.0",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.2.0",
"date": "2026-10-01T00:58:01Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.2.0",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.1.6",
"date": "2026-09-30T18:16:06Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.6",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.1.4",
"date": "2026-09-30T15:04:30Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.4",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.1.3",
"date": "2026-09-30T14:55:17Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.3",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.1.2",
"date": "2026-09-30T11:06:53Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.2",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.1.1",
"date": "2026-09-30T07:40:42Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.1",
"body": "See the assets to download the installer. Existing installations update themselves."
},
{
"tag": "v0.1.0",
"date": "2026-09-30T07:03:22Z",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.0",
"body": "See the assets to download the installer. Existing installations update themselves."
}
]
+628
View File
@@ -0,0 +1,628 @@
{
"schema": 1,
"component": "manager",
"repository": "Corfirean/coa-server-manager",
"consumedFragments": [
"Corfirean/coa-server-manager:20261002-collections-tab-with-transmog-vanity-riding-and.md",
"Corfirean/coa-server-manager:20261002-companion-bots-are-recognised-from-their-default.md",
"Corfirean/coa-server-manager:20261002-delete-an-account-from-the-players-page.md",
"Corfirean/coa-server-manager:20261002-importing-a-server-suggests-the-right-folder-whe.md",
"Corfirean/coa-server-manager:20261002-modules-page-with-cards-status-badges-and-per-mo.md",
"Corfirean/coa-server-manager:20261002-report-a-problem-collects-more-and-ignores-repea.md",
"Corfirean/coa-server-manager:20261002-server-settings-in-one-panel-with-search-tabs-an.md",
"Corfirean/coa-server-manager:20261002-the-update-check-says-you-are-up-to-date-instead.md",
"Corfirean/coa-server-manager:20261003-client-realm-sync.md",
"Corfirean/coa-server-manager:20261003-config-snapshot-restore.md",
"Corfirean/coa-server-manager:20261003-imported-servers-get-the-module-config-files-the.md",
"Corfirean/coa-server-manager:20261003-linux-preview.md",
"Corfirean/coa-server-manager:20261003-modules-marked-soon-stay-greyed-out.md",
"Corfirean/coa-server-manager:20261003-report-browser.md",
"Corfirean/coa-server-manager:20261003-simplified-chinese.md",
"Corfirean/coa-server-manager:20261003-the-bots-page-is-hidden-when-the-module-is-not-o.md",
"Corfirean/coa-server-manager:20261003-updating-the-game-client-shows-a-progress-bar-on.md",
"Corfirean/coa-server-manager:20261003-wildcard-realm-profiles.md",
"Corfirean/coa-server-manager:20261004-auction-item-types.md",
"Corfirean/coa-server-manager:20261004-fractional-settings.md",
"Corfirean/coa-server-manager:20261004-independent-client-launch.md",
"Corfirean/coa-server-manager:20261004-module-settings.md",
"Corfirean/coa-server-manager:20261004-recover-server-updates.md",
"Corfirean/coa-server-manager:20261004-repair-and-database-validation.md",
"Corfirean/coa-server-manager:20261004-scaling-lfg-leech-settings.md",
"Corfirean/coa-server-manager:20261004-simultaneous-realms.md",
"Corfirean/coa-server-manager:20261004-squid-playerbots.md",
"Corfirean/coa-server-manager:20261005-connect-without-server.md",
"Corfirean/coa-server-manager:20261005-imported-squid-warning.md",
"Corfirean/coa-server-manager:20261005-launch-without-updating.md",
"Corfirean/coa-server-manager:20261005-linux-docker-server-installer.md",
"Corfirean/coa-server-manager:20261005-linux-docker-server-settings.md",
"Corfirean/coa-server-manager:20261005-linux-install-screen-and-client.md",
"Corfirean/coa-server-manager:20261005-retry-update-startup.md",
"Corfirean/coa-server-manager:20261005-squid-release-metadata.md",
"Corfirean/coa-server-manager:20261005-squid-version-reports.md",
"Corfirean/coa-server-manager:20261005-update-history-hotfix.md",
"Corfirean/coa-server-manager:20261005-update-legacy-realm-launcher.md",
"Corfirean/coa-server-manager:20261005-validate-release-database.md",
"Corfirean/coa-server-manager:20261006-ah-bot-needs-no-character-guid.md",
"Corfirean/coa-server-manager:20261006-lan-address-selector.md",
"Corfirean/coa-server-manager:20261006-npc-enchanter-placement-hint.md",
"Corfirean/coa-server-manager:20261006-prevent-server-downgrades.md",
"Corfirean/coa-server-manager:20261006-recognize-managed-launcher.md",
"Corfirean/coa-server-manager:20261006-squid-first-start-timeout.md",
"Corfirean/coa-server-manager:20261006-update-preflight-and-recovery-guards.md",
"Corfirean/coa-server-manager:20261006-update-sql-and-status-guards.md",
"Corfirean/coa-server-manager:20261007-diagnostics-valid-json.md",
"Corfirean/coa-server-manager:20261007-one-click-diagnostics.md",
"Corfirean/coa-server-manager:20261007-report-routing.md",
"Corfirean/coa-server-manager:20261008-backup-folder-and-bot-choices.md",
"Corfirean/coa-server-manager:20261008-bot-settings-drop-down.md",
"Corfirean/coa-server-manager:20261008-dashboard-delivery.md",
"Corfirean/coa-server-manager:20261008-database-error-not-misread.md",
"Corfirean/coa-server-manager:20261008-hide-squid-bot-accounts.md",
"Corfirean/coa-server-manager:20261008-squid-dashboard-tab.md",
"Corfirean/coa-server-manager:20261008-start-after-update-options.md",
"Corfirean/coa-server-manager:20261008-update-check-survives-damaged-state.md",
"Corfirean/coa-server-manager:20261008-wildcard-repair-checksum.md",
"Corfirean/coa-server-manager:20261008-world-server-from-coa-bots-folder.md",
"Corfirean/coa-server-manager:20261010-automatic-update-recovery.md",
"Corfirean/coa-server-manager:20261010-complete-mysql-update-recovery.md",
"Corfirean/coa-server-manager:20261010-custom-races-bot-exclusions.md",
"Corfirean/coa-server-manager:20261010-custom-races-update.md",
"Corfirean/coa-server-manager:20261010-custom-races.md",
"Corfirean/coa-server-manager:20261010-database-recovery-diagnostics.md",
"Corfirean/coa-server-manager:20261010-headless-world-startup.md",
"Corfirean/coa-server-manager:20261010-immutable-update-channels.md",
"Corfirean/coa-server-manager:20261010-isolated-update-validation.md",
"Corfirean/coa-server-manager:20261010-local-host-while-sharing.md",
"Corfirean/coa-server-manager:20261010-missing-database-defaults.md",
"Corfirean/coa-server-manager:20261010-module-base-update-replay.md",
"Corfirean/coa-server-manager:20261010-private-database-update-rehearsal.md",
"Corfirean/coa-server-manager:20261010-release-compatibility.md",
"Corfirean/coa-server-manager:20261010-schema-validation-report.md",
"Corfirean/coa-server-manager:20261010-settings-blank-window.md",
"Corfirean/coa-server-manager:20261010-unsaved-changes-and-crash-recovery.md",
"Corfirean/coa-server-manager:20261011-player-race-patches.md",
"Corfirean/coa-server-manager:20261012-bulk-navigation-update.md",
"Corfirean/coa-server-manager:20261012-rehearsal-path-normalization.md",
"Corfirean/coa-server-manager:20261013-custom-races-client-update.md",
"Corfirean/coa-server-manager:20261013-docker-remote-client-restart.md",
"Corfirean/coa-server-manager:20261013-installed-module-versions.md",
"Corfirean/coa-server-manager:20261013-long-rehearsal-paths.md",
"Corfirean/coa-server-manager:20261013-maintenance-diagnostic-export.md",
"Corfirean/coa-server-manager:20261013-preserve-client-patches.md",
"Corfirean/coa-server-manager:20261013-recheck-installed-release.md",
"Corfirean/coa-server-manager:20261013-startup-error-details.md",
"Corfirean/coa-server-manager:20261013-stock-bot-races.md",
"Corfirean/coa-server-manager:20261013-update-check-start-contention.md",
"Corfirean/coa-server-manager:20261013-upstream-bot-race-preset.md",
"Corfirean/coa-server-manager:20261014-already-running-mysql-is-recognized-before-start.md",
"Corfirean/coa-server-manager:20261014-current-clean-install.md",
"Corfirean/coa-server-manager:20261014-custom-races-grummle-client.md",
"Corfirean/coa-server-manager:20261014-failed-repairs-restore-the-previous-server-after.md",
"Corfirean/coa-server-manager:20261014-overview-update-errors.md",
"Corfirean/coa-server-manager:20261014-private-update-checks-show-preparation-stages-an.md",
"Corfirean/coa-server-manager:20261014-repair-recovery-copies-preserve-all-bundled-data.md",
"Corfirean/coa-server-manager:20261014-server-checks-report-interrupted-squid-database.md",
"Corfirean/coa-server-manager:20261014-shutdown-remains-available-when-startup-template.md",
"Corfirean/coa-server-manager:20261014-successful-server-repairs-no-longer-block-later.md",
"Corfirean/coa-server-manager:20261016-consistent-update-migrations.md",
"Corfirean/coa-server-manager:20261016-direct-database-upgrades.md",
"Corfirean/coa-server-manager:20261016-idempotent-validation-stop.md",
"Corfirean/coa-server-manager:20261016-intermediate-updates.md",
"Corfirean/coa-server-manager:20261016-lightweight-update-checks.md",
"Corfirean/coa-server-manager:20261016-preserve-core-migration-history.md",
"Corfirean/coa-server-manager:20261016-shared-auth-migration-history.md",
"Corfirean/coa-server-manager:20261016-update-compatibility-wording.md",
"Corfirean/coa-server-manager:20261018-custom-races-client.md",
"Corfirean/coa-server-manager:20261019-portable-stop-diagnostics.md",
"Corfirean/coa-server-manager:20261019-remote-character-class-names.md",
"Corfirean/coa-server-manager:20261020-client-elevation.md",
"Corfirean/coa-server-manager:20261021-custom-races-switching.md",
"Corfirean/coa-server-manager:20261021-portable-experimental.md",
"Corfirean/coa-server-manager:20261021-scaling-health-settings.md",
"Corfirean/coa-server-manager:20261021-warcraft-reborn-realm.md",
"Corfirean/coa-server-manager:20261022-custom-races-client.md"
],
"releases": [
{
"tag": "v0.1.0",
"version": "0.1.0",
"date": "2026-09-30",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.0",
"commit": "0a9be3d80030f8549a2925e176ab0ee829603c44",
"notes": [
"Initial Windows release: install or import a server and manage startup from one app",
"Configure server and bot settings, manage accounts, create backups and restore saved data",
"Launch the game client, set up sharing with friends and collect diagnostics"
]
},
{
"tag": "v0.1.1",
"version": "0.1.1",
"date": "2026-09-30",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.1",
"commit": "f620f87432e5affb84bdc14ca0bb0f5a8bb5c534",
"notes": [
"Added translations throughout setup, server updates, backups, diagnostics and settings",
"Added clearer errors when server packages are unavailable or a server cannot be reached"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.1.0...v0.1.1"
},
{
"tag": "v0.1.2",
"version": "0.1.2",
"date": "2026-09-30",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.2",
"commit": "0e1016e93c457c8aeea3ba31416be66a201a19ef",
"notes": [
"Manager updates download in the background and install when the app closes",
"Missing module configuration files are created from the supplied defaults"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.1.1...v0.1.2"
},
{
"tag": "v0.1.3",
"version": "0.1.3",
"date": "2026-09-30",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.3",
"commit": "4eda03ab6355a6cf280865bca91ce8d2fc6167e7",
"notes": [
"The interface scales with the window",
"Each friend connection mode shows the realmlist line to use"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.1.2...v0.1.3"
},
{
"tag": "v0.1.4",
"version": "0.1.4",
"date": "2026-09-30",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.4",
"commit": "db16d95558910d4cdc97edbce97ec78aaf535a83",
"notes": [
"Sharing settings are reapplied before server startup",
"Added a restart shortcut after enabling play with friends"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.1.3...v0.1.4"
},
{
"tag": "v0.1.6",
"version": "0.1.6",
"date": "2026-09-30",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.1.6",
"commit": "f7d29b09efd54ab3a0f830d4d6891013038398e4",
"notes": [
"Overview shows world speed",
"Improved companion setup, including requests made while the server is stopped",
"Expanded translated settings and connection guides"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.1.4...v0.1.6"
},
{
"tag": "v0.2.0",
"version": "0.2.0",
"date": "2026-10-01",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.2.0",
"commit": "0940e9cf28759a9489d53dbd28882d6b46cb0076",
"notes": [
"Added companion creation progress and controls to stop creation, take bots offline or remove them",
"Choose how many bots log in when the server starts",
"Added a control to log out a chosen number of bots and reduce server load"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.1.6...v0.2.0"
},
{
"tag": "v0.3.0",
"version": "0.3.0",
"date": "2026-10-01",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.3.0",
"commit": "c8f6d61b98ceb0920ba605467eb73b1364d24404",
"notes": [
"Download and update the game client through the Manager while preserving custom files",
"Server updates are checked automatically and offered on Overview"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.2.0...v0.3.0"
},
{
"tag": "v0.3.1",
"version": "0.3.1",
"date": "2026-10-01",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.3.1",
"commit": "d01e401f3ce5c5f9a7daeb721815e2cf5fe1d694",
"notes": [
"Added named realmlists and a switcher beside Play",
"Added account management and a prefilled GitHub problem-report form",
"Server updates continue when leaving the settings page"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.3.0...v0.3.1"
},
{
"tag": "v0.4.0",
"version": "0.4.0",
"date": "2026-10-02",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.4.0",
"commit": "53aa117ec2e2f4c0572a74829c41e99d5a4eedb6",
"notes": [
"Added module cards with switches and settings",
"Added searchable server settings and appearance and collection controls",
"Added account deletion with confirmation",
"Fixed update checks offering an already installed version"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.3.1...v0.4.0"
},
{
"tag": "v0.5.0",
"version": "0.5.0",
"date": "2026-10-03",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.5.0",
"commit": "1bcef5f8c2e85b28eb466e720002d2fbf2b67161",
"notes": [
"Imported servers get the module config files they lack",
"Modules marked Soon stay greyed out",
"The Bots page is hidden when the module is not on the server",
"Updating the game client shows a progress bar on its button",
"Switch between CoA and Wildcard while keeping separate characters and module settings"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.4.0...v0.5.0"
},
{
"tag": "v0.6.0",
"version": "0.6.0",
"date": "2026-10-03",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.0",
"commit": "162bfd1e04ca4faa205b5c38432f56b07066c6be",
"notes": [
"The game client follows the realm selected in the Manager",
"Configuration snapshots no longer overwrite each other",
"Experimental Linux startup and Docker server controls",
"Problem reports open in the default browser",
"Simplified Chinese is available in the Manager"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.5.0...v0.6.0"
},
{
"tag": "v0.6.1",
"version": "0.6.1",
"date": "2026-10-04",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.1",
"commit": "2e3fb24e1af5c3ee9a787c345cbcce973739fe69",
"notes": [
"Server multipliers now accept decimals typed with a dot or comma",
"Play launches only the game client without starting your local server",
"Check database structure and repair server files with a safety backup",
"Start CoA and Wildcard together from Settings"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.0...v0.6.1"
},
{
"tag": "v0.6.2",
"version": "0.6.2",
"date": "2026-10-04",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.2",
"commit": "b90b377fc308210abdf7fc63ea2f2ac1eac29f4a",
"notes": [
"Choose which item types the auction bot sells",
"Module settings now offer translated controls and an enemy damage slider",
"Failed server updates restore databases and files together",
"Configure SQUID Playerbots and prevent conflicting bot systems",
"Server packages must include a complete database schema check"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.1...v0.6.2"
},
{
"tag": "v0.6.3",
"version": "0.6.3",
"date": "2026-10-04",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.3",
"commit": "07ec22798f2b81f33e470837ece7fec693ac24bb",
"notes": [
"Configure partial dungeon parties and open-world life steal in module settings",
"Download and update the game client without installing a server"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.2...v0.6.3"
},
{
"tag": "v0.6.4",
"version": "0.6.4",
"date": "2026-10-05",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.4",
"commit": "6c771fae20ac59870f5990a6495f8cd035caab65",
"notes": [
"Start your installed server or client without downloading an available update",
"Finish a server update that was blocked by a startup failure"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.3...v0.6.4"
},
{
"tag": "v0.6.5",
"version": "0.6.5",
"date": "2026-10-06",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.5",
"commit": "277ebba847912e113657bcef4ed53e4e4e708161",
"notes": [
"Imported SquidBots servers warn before replacing their original binary",
"SquidBots settings follow the options shipped by the module",
"SquidBots shows its bundled release and includes it in problem reports",
"Server updates accept verified published migration variants without rerunning them",
"Server updates replace older launchers adapted by the Manager for Wildcard",
"Older server packages cannot replace a newer installation",
"Checking again no longer offers an already installed launcher update",
"Large SQUID database imports have more time to finish on first startup",
"Update history conflicts are detected before server files change",
"Pending database updates remain visible when the server version matches"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.4...v0.6.5"
},
{
"tag": "v0.6.6",
"version": "0.6.6",
"date": "2026-10-07",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.6",
"commit": "fac6ce469a77c134437484a78e91e7811df8a389",
"notes": [
"Install a server from a package on Linux, running in Docker (experimental)",
"The Auction House Bot card no longer asks for a character GUID",
"Choose and remember the local network address friends use to connect",
"The NPC Enchanter card explains how to place and remove the NPC",
"Diagnostic files with database checks stay readable after secrets are removed",
"One button collects all diagnostics, including crashes and Windows crash records",
"Problem reports go to the right project"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.5...v0.6.6"
},
{
"tag": "v0.6.7",
"version": "0.6.7",
"date": "2026-10-08",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.7",
"commit": "7540527d3035660f1c45ea66497e0412cc65249c",
"notes": [
"Choose where backups are stored, for example on another drive",
"Bot settings with a fixed list of values show a drop-down",
"The bot dashboard loads completely and can be opened in the browser",
"Accounts of the SQUID bots are hidden from the Accounts list",
"A Dashboard tab shows what the SQUID bots are doing",
"A stopped server stays stopped after an update, unless you ask it to start",
"Checking for updates no longer fails with \"Something went wrong\" after a crash",
"Server updates no longer stop with \"has a different checksum\" on the Wildcard repair",
"A world server started from the coa-bots folder is recognised"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.6...v0.6.7"
},
{
"tag": "v0.6.8",
"version": "0.6.8",
"date": "2026-10-10",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.8",
"commit": "ee59708fbfcdb5f4e30c62f2e4aa952f82604cac",
"notes": [
"Linux: install from the install screen, play with Wine or Proton (experimental)",
"A failing database update is no longer reported as \"database not running",
"Unsaved settings are protected and crashed servers can restart automatically"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.7...v0.6.8"
},
{
"tag": "v0.6.9",
"version": "0.6.9",
"date": "2026-10-10",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.9",
"commit": "374a3f9c2322f30900f7f8a0549a643c8a511567",
"notes": [
"Custom races can be enabled and their client patches installed from Modules"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.8...v0.6.9"
},
{
"tag": "v0.6.10",
"version": "0.6.10",
"date": "2026-10-10",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.10",
"commit": "b0951dbee7516334d8ae3baa1c7bf7635f6cb202",
"notes": [
"Custom race client patches update to the latest races and appearances"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.9...v0.6.10"
},
{
"tag": "v0.6.11",
"version": "0.6.11",
"date": "2026-10-10",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.11",
"commit": "376f1dbb469ece197aa1caf68b3e5fdd73e3b277",
"notes": [
"Server and Bots settings open without a blank window"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.10...v0.6.11"
},
{
"tag": "v0.6.12",
"version": "0.6.12",
"date": "2026-10-10",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.12",
"commit": "94d4ade1c73c6a64018dc182fe3a3ba87607d7a2",
"notes": [
"Background world servers no longer stop because the local console is closed"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.11...v0.6.12"
},
{
"tag": "v0.6.13",
"version": "0.6.13",
"date": "2026-10-11",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.13",
"commit": "ed7bf7262bc43de109f17a2812bc41c34d99c51d",
"notes": [
"Failed server startup after an update automatically restores the previous version",
"Windows update recovery preserves database procedures triggers and views",
"Custom races bot setting synchronizes Playerbots excluded race lists",
"Blocked updates identify database objects that prevent automatic recovery",
"Updates can follow signed channels that select a complete server release",
"Server updates check startup privately before allowing players to connect",
"Updates repair missing database defaults from the verified release schema",
"Server updates no longer replay module installation scripts on existing databases",
"Windows server updates are tested on a private copy before changing your installation",
"Server updates can require a tested combination of platform and installed version",
"Failed database checks keep a complete report after rollback",
"Players can install custom race patches when connecting to a friend"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.12...v0.6.13"
},
{
"tag": "v0.6.14",
"version": "0.6.14",
"date": "2026-10-11",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.14",
"commit": "58c976a388215bea8bbf232503810c6519a33562",
"notes": [
"Fixed server update checks failing with long Windows paths"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.13...v0.6.14"
},
{
"tag": "v0.6.15",
"version": "0.6.15",
"date": "2026-10-11",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.15",
"commit": "d86a1676fed98f9e4556617a603959e42bc401c8",
"notes": [
"Large server updates no longer repeatedly rewrite the recovery journal"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.14...v0.6.15"
},
{
"tag": "v0.6.16",
"version": "0.6.16",
"date": "2026-10-13",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.16",
"commit": "b45e2502ece1a5578555d7d5dfe36a95b374a58e",
"notes": [
"Linux Docker servers follow Settings, sharing and backups (experimental)",
"Local hosts can connect while sharing a server online",
"Custom race client installation follows the current upstream release",
"Docker friend mode requests a restart when remote client support changes",
"Module cards show installed server revisions and the client race patch version",
"Server updates accept long Windows paths inside the private test server",
"Diagnostics record blocked startup requests and maintenance owner information",
"Client race patches can be repaired and removed while preserving original files",
"Checking an already installed server update no longer reports incompatibility",
"Failed server startup shows service errors even when the launcher reported ready",
"Starting the server waits for update checks and identifies ongoing maintenance"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.15...v0.6.16"
},
{
"tag": "v0.6.17",
"version": "0.6.17",
"date": "2026-10-15",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.17",
"commit": "dc323df681ca77fb15ce4284da0feadfe83e0c7e",
"notes": [
"Already running MySQL is recognized before startup readiness checks",
"New servers install the current stable build directly",
"Install the matching client patches for Grummle",
"Failed repairs restore the previous server after final verification errors",
"Failed server updates show an error on the Overview screen",
"Private update checks show preparation stages and stop failure details",
"Repair recovery copies preserve all bundled database objects and accounts",
"Server checks report interrupted SQUID database imports",
"Shutdown remains available when startup templates or database records are damaged",
"Successful server repairs no longer block later server actions"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.16...v0.6.17"
},
{
"tag": "v0.6.18",
"version": "0.6.18",
"date": "2026-10-16",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.18",
"commit": "16396407ad63c1b596dc4fd7e2672ef87cf23f56",
"notes": [
"Older servers find a tested intermediate update instead of being blocked"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.17...v0.6.18"
},
{
"tag": "v0.6.19",
"version": "0.6.19",
"date": "2026-10-16",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.19",
"commit": "cda7875a9a2061d4b576912cc6943220ff6fddba",
"notes": [
"Update checks no longer download a server archive to compare its launcher"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.18...v0.6.19"
},
{
"tag": "v0.6.21",
"version": "0.6.21",
"date": "2026-10-16",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.21",
"commit": "e3fbb2e4b6dbe40fa00264abcd12b4cfd9702305",
"notes": [
"Server updates apply migrations once after saving a recovery point",
"Older server versions update directly without installing intermediate packages",
"Update validation recognizes an already stopped server",
"Recorded core SQL stays applied after old history tables are cleaned up",
"Switching realms no longer repeats shared account database updates",
"Update offers distinguish version support from database validation"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.19...v0.6.21"
},
{
"tag": "v0.6.22",
"version": "0.6.22",
"date": "2026-10-19",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.6.22",
"commit": "65df6604846d5492e40a6b08a797199478666e24",
"notes": [
"Updated Custom Races client patches to 1.8",
"Included the updated race appearance and cursor addons",
"Existing client files are preserved when installing or removing race patches"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.21...v0.6.22"
},
{
"tag": "v0.7.0",
"version": "0.7.0",
"date": "2026-10-21",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.7.0",
"notes": [
"Keep portable characters and their saved versions in the Manager, and use them on compatible realms",
"Make an existing character portable from a connected realm",
"Browse public realms and see their connection details",
"Link an existing realm account or create one on hosts that allow it",
"Switch between CoA, Wildcard and Warcraft Reborn realm profiles on supported server builds",
"Adjust creature health for different locations and enemy types in Content Scaling",
"Switch Custom Races patches faster, with visible feedback while changes are applied",
"Handle Windows requests for administrator permission when launching the game"
],
"commit": "d22ff09ac6fd7628aac14d90ec23f547cd9b4bf1",
"experimentalFeatures": [
"portable-characters"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.6.22...v0.7.0",
"notice": "Portable transfers into Warcraft Reborn are not supported yet.",
"links": [
"[Player guides and screenshots](https://github.com/Corfirean/coa-server-manager/tree/d22ff09ac6fd7628aac14d90ec23f547cd9b4bf1/docs)"
]
},
{
"tag": "v0.7.1",
"version": "0.7.1",
"date": "2026-10-22",
"url": "https://github.com/Corfirean/coa-server-manager/releases/tag/v0.7.1",
"notes": [
"Updated Custom Races client patches to the complete 1.8.3 package",
"Race patch installation now includes the hide-armor and talent-reset addons"
],
"commit": "8d2befd7f08532b2cb9e0bf2f973bb631f3890e5",
"experimentalFeatures": [
"portable-characters"
],
"compare": "https://github.com/Corfirean/coa-server-manager/compare/v0.7.0...v0.7.1",
"notice": "Close the game, then open Modules → Custom Races → Repair Client Patch to update existing patches, or Install Client Patch for a new installation. Updating the Manager does not automatically replace game files or update your server.",
"links": [
"[Guides and screenshots](https://github.com/Corfirean/coa-server-manager/blob/master/docs/README.md)"
]
}
]
}
@@ -0,0 +1,57 @@
//! Read-only audit, or explicit SQL checks on a disposable database fixture.
use coa_core::{db::{Account, Db}, migrations::{self, Store}, manifest::Manifest, realms::Mode};
fn main() -> Result<(), Box<dyn std::error::Error>> {
let args: Vec<_> = std::env::args().collect();
if args.len() < 3 { return Err("usage: migration_plan <root> <signed-manifest> [--apply|--sql-only <migration-directory>]".into()); }
let root = std::path::Path::new(&args[1]);
let bytes = std::fs::read(&args[2])?;
let signature = std::fs::read_to_string(format!("{}.sig",args[2]))?;
coa_core::signing::verify(&bytes, &signature, coa_core::signing::EMBEDDED_PUBLIC_KEY)?;
let manifest = Manifest::parse(&bytes)?;
for (index,mode) in [Mode::Coa, Mode::Wildcard].into_iter().enumerate() {
let db = Db::from_repack(root, Account::Admin)?.for_realm(mode);
let rows = db.load_readonly()?;
let relevant: Vec<_> = rows.iter().filter(|r| r.id == "rev_20260903_01_live_class_baseline").map(|r| format!("{} {:?} {}",r.db,r.status,r.baseline)).collect();
println!("{} baseline history: {:?}",mode.name(),relevant);
let list: Vec<_> = manifest.migrations.iter().filter(|m|index==0 || m.db!="auth").cloned().collect();
let plan = migrations::status(&db,&list)?;
println!("{} pending: {:?}",mode.name(),plan.iter().filter(|m|m.status!=migrations::Status::Applied).map(|m|(&m.db,&m.id,m.status)).collect::<Vec<_>>());
}
if args.get(3).is_some_and(|s| s == "--apply" || s == "--sql-only") {
assert!(root.file_name().unwrap().to_string_lossy().starts_with("update-database-diagnosis-") && root.join(".coa-update-fixture").is_file(), "Only disposable fixture roots may be changed");
if args[3] == "--sql-only" {
let db = Db::from_repack(root, Account::Admin)?.for_realm(Mode::Coa);
let identity = || -> coa_core::Result<String> {
let mut rows = String::new();
for (schema,table,columns) in [("acore_auth","account","id,HEX(username)"),("acore_characters","characters","guid,account,HEX(name)"),("acore_characters_wildcard","characters","guid,account,HEX(name)")] {
rows.push_str(schema);
rows.push_str(&db.query(&format!("SELECT {columns} FROM {schema}.{table} ORDER BY 1;"))?);
}
Ok(coa_core::fsx::sha256_bytes(rows.as_bytes()))
};
let before = identity()?;
migrations::verify_files(&manifest.migrations,std::path::Path::new(&args[4]))?;
for (index,mode) in [Mode::Coa,Mode::Wildcard].into_iter().enumerate() {
let realm_db = db.clone().for_realm(mode);
let list: Vec<_> = manifest.migrations.iter().filter(|m|index==0 || m.db!="auth").cloned().collect();
let result = migrations::apply_pending(&realm_db,&list,std::path::Path::new(&args[4]),&||Ok("disposable fixture".into()))?;
println!("{} SQL result: {}",mode.name(),serde_json::to_string(&result)?);
if result.failed.is_some() {return Err("SQL failed; fixture preserved".into());}
}
assert_eq!(before,identity()?,"Account or character identity changed");
println!("Account and character identities preserved: {}",before);
let capture = root.join("captured-schema");
coa_core::schema_check::capture(&db,&capture)?;
println!("Captured schema SHA256: {}",coa_core::fsx::sha256_file(&capture.join(coa_core::schema_check::CONTRACT))?);
return Ok(());
}
if args.len() != 5 { return Err("Migration directory required".into()); }
use coa_core::update::{Env, RepackEnv};
let meta = coa_core::registry::metadata_dir_for(root)?;
let env = RepackEnv { root, meta_dir: &meta };
let result = env.migrate(&manifest, std::path::Path::new(&args[4]));
let _ = coa_core::driver::run(root, coa_core::driver::Verb::StopAll);
println!("Application result: {}",serde_json::to_string(&result?)?);
}
Ok(())
}
@@ -0,0 +1,28 @@
//! Exercise the production updater only on a disposable copy of an installed server.
use coa_core::{registry::MetaDir, update::{self, Env, RepackEnv}};
fn main() -> Result<(),Box<dyn std::error::Error>> {
let args:Vec<_>=std::env::args().collect();
if args.len()!=3 && args.len()!=4 {return Err("usage: upgrade_fixture <disposable-root> <signed-package-directory>".into());}
let root=std::path::Path::new(&args[1]);
assert!(root.file_name().unwrap().to_string_lossy().starts_with("coa-schema-fixture-real") && root.join(".coa-update-fixture").is_file(),"Disposable fixture required");
let meta=coa_core::registry::metadata_dir_for(root)?;
let (_,installed)=MetaDir::open(&meta)?;
assert_eq!(coa_core::fsx::canonicalize_lenient(std::path::Path::new(&installed.server_path))?,coa_core::fsx::canonicalize_lenient(root)?,"Fixture metadata must point to the copy");
assert!(installed.client_path.is_none(),"Detach the real client from fixture metadata");
let env=RepackEnv{root,meta_dir:&meta};
if args.get(3).is_some_and(|flag|flag=="--validate-only") {
let result=env.validate();
let _=coa_core::driver::run(root,coa_core::driver::Verb::StopAll);
result?;
println!("Startup and identity validation passed");
return Ok(());
}
let outcome=update::apply(&update::Params{root,meta_dir:&meta,source:coa_core::pkgsource::Source::Dir(args[2].clone().into()),trusted_key:coa_core::signing::EMBEDDED_PUBLIC_KEY,cancel:Default::default(),resolutions:Default::default(),env:&env,fail_after_ops:None},&|phase,percent|println!("{percent}% {phase}"));
let stopped=coa_core::driver::run(root,coa_core::driver::Verb::StopAll);
println!("Shutdown after test: {:?}",stopped.map(|r|r.ok));
let outcome=outcome?;
println!("Final transaction: {:?}",outcome.txn.state);
env.ensure_stopped()?;
assert_eq!(outcome.txn.state,update::State::Committed);
Ok(())
}
+2 -2
View File
@@ -212,7 +212,7 @@ pub fn config_files(root: &Path) -> Vec<String> {
for e in rd.flatten() {
let n = e.file_name().to_string_lossy().to_string();
// database.json holds generated database passwords; it is launcher state, not configuration.
if e.path().is_file() && (n.contains(".template") || n == "repack.json") {
if e.path().is_file() && (n.contains(".template") || n == "repack.json" || n == "docker.json") {
out.push(format!("Settings/{n}"));
}
}
@@ -336,7 +336,7 @@ pub fn create(root: &Path, meta: &Path, kind: Kind, trigger: Trigger, label: Opt
return Err(e);
}
};
let mysql_snapshot = if kind == Kind::Full && trigger == Trigger::BeforeUpdate && cfg!(windows) && !crate::docker::is_docker(root) {
let mysql_snapshot = if kind == Kind::Full && matches!(trigger, Trigger::BeforeUpdate | Trigger::BeforeRepair) && cfg!(windows) && !crate::docker::is_docker(root) {
progress("Saving a complete stopped MySQL recovery copy");
match crate::mysql_snapshot::capture(root, &partial) {
Ok(snapshot) => Some(snapshot),
+1 -1
View File
@@ -24,7 +24,7 @@ pub struct Params<'a> {
}
/// Auth tables that hold player data. Everything else (permissions, build info, realmlist) is configuration.
const AUTH_PLAYER_TABLES: [&str; 11] =
pub(crate) const AUTH_PLAYER_TABLES: [&str; 11] =
["account", "account_access", "account_banned", "account_muted", "account_password_history", "ip_banned", "logs", "logs_ip_actions", "realmcharacters", "uptime", "rbac_account_permissions"];
fn copy_dir(src: &Path, dst: &Path, skip: &dyn Fn(&str) -> bool, base: &Path) -> Result<u64> {
+35
View File
@@ -206,6 +206,10 @@ pub fn drift(root: &Path) -> Result<Vec<String>> {
}
pub fn load(root: &Path, scope: Scope) -> Result<SettingsView> {
// A Docker server that was added but never started has no active world configuration yet; create it from its template.
if scope == Scope::Server && crate::docker::is_docker(root) {
let _ = crate::docker::ensure_main_configs(root);
}
let schema = scope.schema();
let t = targets(root, scope)?;
let conf = load_conf(&t.read)?;
@@ -213,9 +217,12 @@ pub fn load(root: &Path, scope: Scope) -> Result<SettingsView> {
let known: BTreeSet<&str> = schema.settings.iter().map(|s| s.key.as_str()).collect();
let unknown_keys = conf.entries().filter(|(k, _)| !known.contains(k)).count();
// On a Docker server some settings are fixed by the backend (environment beats the file): not offered.
let docker = scope == Scope::Server && crate::docker::is_docker(root);
let settings = schema
.settings
.iter()
.filter(|s| !(docker && crate::docker::managed_setting(&s.key)))
.map(|s| {
let raw = conf.get(&s.key);
let (value, problem) = match raw {
@@ -286,11 +293,19 @@ pub fn validate(root: &Path, scope: Scope, changes: &BTreeMap<String, Value>) ->
let schema = scope.schema();
let mut errors = Vec::new();
let mut raws = BTreeMap::new();
let docker = scope == Scope::Server && crate::docker::is_docker(root);
for (key, value) in changes {
match schema.get(key) {
None => errors.push(FieldError { key: key.clone(), message: "is not a setting the Manager can change".into() }),
Some(_) if docker && crate::docker::managed_setting(key) => {
errors.push(FieldError { key: key.clone(), message: "is set by the Docker backend; changing it in the file would have no effect".into() })
}
Some(s) => match s.to_raw(value) {
Ok(raw) => {
if let Some(problem) = crate::docker::setting_problem(key, &raw).filter(|_| crate::docker::is_docker(root)) {
errors.push(FieldError { key: key.clone(), message: problem.into() });
continue;
}
raws.insert(key.clone(), raw);
}
Err(m) => errors.push(FieldError { key: key.clone(), message: m }),
@@ -736,6 +751,26 @@ Spellbook.New = 5
assert!(list_snapshots(&meta).is_empty(), "no snapshot for a rejected save");
}
#[test]
fn a_docker_server_is_not_offered_the_settings_its_backend_fixes() {
let (_d, root, meta) = fixture();
let keys = |root: &Path| load(root, Scope::Server).unwrap().settings.into_iter().map(|s| s.meta.key).collect::<Vec<_>>();
for k in ["Ra.Enable", "Ra.IP", "Updates.EnableDatabases", "PlayerLimit"] {
assert!(keys(&root).iter().any(|x| x == k), "a repack offers {k}");
}
fs::write(root.join("Settings/docker.json"), br#"{"project":"t1"}"#).unwrap();
let docker = keys(&root);
for k in ["Ra.Enable", "Ra.IP", "Updates.EnableDatabases"] {
assert!(!docker.iter().any(|x| x == k), "a Docker server does not offer {k}: the containers' environment beats the file");
}
assert!(docker.iter().any(|x| x == "PlayerLimit"), "everything else is still offered");
// Forcing a change through is refused and says why, instead of being saved and ignored.
let err = save(&root, &meta, Scope::Server, &set(&[("Ra.Enable", json!(0))])).unwrap_err().to_string();
assert!(err.contains("Docker backend"), "{err}");
assert!(!fs::read_to_string(root.join("Core/configs/worldserver.conf")).unwrap().contains("Ra.Enable = 0"));
save(&root, &meta, Scope::Server, &set(&[("PlayerLimit", json!(77))])).unwrap();
}
#[test]
fn missing_key_is_appended_and_noop_saves_do_nothing() {
let (_d, root, meta) = fixture();
+13
View File
@@ -51,6 +51,19 @@ fn classify(l: &str) -> Level {
}
}
/// The last `max_lines` lines of some text (after filtering), classified and redacted like the lines of a log file.
pub fn lines_from_text(text: &str, filter: Option<&str>, max_lines: usize) -> Vec<Line> {
let needle = filter.map(str::to_lowercase).filter(|n| !n.is_empty());
let out: Vec<Line> = text
.lines()
.filter(|l| !l.trim().is_empty())
.filter(|l| needle.as_ref().map(|n| l.to_lowercase().contains(n)).unwrap_or(true))
.map(|l| Line { level: classify(l), text: crate::diag::redact(l) })
.collect();
let skip = out.len().saturating_sub(max_lines);
out.into_iter().skip(skip).collect()
}
/// The last `max_lines` lines (after filtering) from the last ~512 KB of a log. Works on multi-gigabyte files.
pub fn tail(path: &Path, filter: Option<&str>, max_lines: usize) -> Result<Vec<Line>> {
const WINDOW: u64 = 512 * 1024;
+74 -16
View File
@@ -14,20 +14,24 @@ use crate::download::Cancel;
use crate::error::{Error, Result};
use crate::fsx;
pub const CLIENT_VERSION: &str = "1.5";
pub const CLIENT_VERSION: &str = "1.7";
pub const MANIFEST_FILENAME: &str = "custom_races_patch_manifest.txt";
pub const ARCHIVES: &[(&str, &str)] = &[
(
"CoA-Custom-1.5-main-20261009-cb2a0ca4.zip",
"https://github.com/ilusixn/azerothcore-wotlk-coa/releases/download/coa-custom-1.5/CoA-Custom-1.5-main-20261009-cb2a0ca4.zip",
"CoA-Custom-1.6-main-20261012-4dd3a4ec.zip",
"https://github.com/ilusixn/azerothcore-wotlk-coa/releases/download/coa-custom-1.6/CoA-Custom-1.6-main-20261012-4dd3a4ec.zip",
),
(
"CoA-Custom-1.5-client-part1.zip",
"https://github.com/ilusixn/azerothcore-wotlk-coa/releases/download/coa-custom-1.5/CoA-Custom-1.5-client-part1.zip",
"CoA-Custom-1.6-client-part1.zip",
"https://github.com/ilusixn/azerothcore-wotlk-coa/releases/download/coa-custom-1.6/CoA-Custom-1.6-client-part1.zip",
),
(
"CoA-Custom-1.5-client-part2.zip",
"https://github.com/ilusixn/azerothcore-wotlk-coa/releases/download/coa-custom-1.5/CoA-Custom-1.5-client-part2.zip",
"CoA-Custom-1.6-client-part2.zip",
"https://github.com/ilusixn/azerothcore-wotlk-coa/releases/download/coa-custom-1.6/CoA-Custom-1.6-client-part2.zip",
),
(
"CoA-Custom-1.7-client.zip",
"https://github.com/ilusixn/azerothcore-wotlk-coa/releases/download/coa-custom-1.7/CoA-Custom-1.7-client.zip",
),
];
@@ -52,13 +56,18 @@ pub const RACE_MPQS: &[&str] = &[
"patch-ZJ.MPQ",
"patch-ZZE.MPQ",
"patch-ZZR.MPQ",
"patch-ZZS.MPQ",
"patch-ZZT.MPQ",
"patch-ZZU.MPQ",
"patch-ZZV.MPQ",
"patch-ZZW.MPQ",
];
pub const EXCLUDED_BOT_RACES_RECOMMENDED: &str =
"19,27,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128";
"9,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128";
pub const EXCLUDED_BOT_RACES_EXPERIMENTAL: &str =
"16,19,27,32,50,52,53,65,66,67,68,69,70,71,72,74,77,80,81";
"65,72,77,80,81";
/// Finds all active playerbots.conf files on this server (root, coa-bots folder, and secondary realm).
pub fn playerbots_conf_paths(root: &Path) -> Vec<PathBuf> {
@@ -99,7 +108,8 @@ pub fn is_bot_custom_races_allowed(root: &Path) -> Option<bool> {
if let Ok(conf) = crate::config::parser::ConfFile::parse_bytes(&bytes) {
if let Some(val) = conf.get("AiPlayerbot.ExcludedBotRaces") {
let clean = val.trim().trim_matches(['"', '\'']);
return Some(!clean.contains("33") && !clean.contains("128"));
let excluded: std::collections::BTreeSet<u32> = clean.split(',').filter_map(|race| race.trim().parse().ok()).collect();
return Some((1..=128).filter(|race| ![1, 2, 3, 4, 5, 6, 7, 8, 10, 11].contains(race)).any(|race| !excluded.contains(&race)));
}
}
}
@@ -146,6 +156,7 @@ pub enum ClientPatchState {
#[serde(rename_all = "camelCase")]
pub struct ClientPatchStatus {
pub state: ClientPatchState,
pub version: Option<String>,
pub installed_mpqs: usize,
pub total_mpqs: usize,
pub has_root_files: bool,
@@ -159,8 +170,9 @@ struct PatchFile {
}
fn patch_files(client: &Path) -> Result<Vec<PatchFile>> {
let bundled: std::collections::BTreeMap<String, String> = serde_json::from_str(include_str!("custom_races_hashes.json"))?;
let manifest = fs::read_to_string(client.join(MANIFEST_FILENAME)).unwrap_or_default();
let hashes = if manifest.lines().any(|line| line == "# CoA Custom 1.5") { include_str!("custom_races_hashes_1_5.json") } else if manifest.lines().any(|line| line == "# CoA Custom 1.6") { include_str!("custom_races_hashes_1_6.json") } else { include_str!("custom_races_hashes.json") };
let bundled: std::collections::BTreeMap<String, String> = serde_json::from_str(hashes)?;
let mut files = std::collections::BTreeMap::new();
for line in manifest.lines() {
let path = if let Some(name) = line.strip_prefix("MPQ:") {
@@ -210,7 +222,8 @@ pub fn client_patch_status(client_path: &Path) -> ClientPatchStatus {
} else if !manifest.is_empty() || patch_files(client_path).is_ok_and(|files| files.iter().filter(|file| file.path.starts_with("Data/") && file.path != "Data/patch-T.MPQ").any(|file| matches_patch(&client_path.join(&file.path), file) || matches_patch(&suffix(&client_path.join(&file.path), "disabled"), file))) {
ClientPatchState::PartiallyInstalled
} else { ClientPatchState::NotInstalled };
ClientPatchStatus { state, installed_mpqs, total_mpqs: RACE_MPQS.len(), has_root_files,
let version = manifest.lines().find_map(|line| line.strip_prefix("# CoA Custom ").map(str::to_string));
ClientPatchStatus { state, version, installed_mpqs, total_mpqs: RACE_MPQS.len(), has_root_files,
client_path: Some(client_path.to_string_lossy().into_owned()) }
}
@@ -358,9 +371,10 @@ pub fn uninstall_client_patches(client: &Path) -> Result<()> {
fn verify_archive(path: &Path, filename: &str) -> Result<()> {
let expected = match filename {
"CoA-Custom-1.5-main-20261009-cb2a0ca4.zip" => "117f5d99add18191676be6648fb0f2e99019a06c18705c71454e371e7df2afd2",
"CoA-Custom-1.5-client-part1.zip" => "19eb372805b3268653914da03377d67f056e5b50d818a8a06a5f260aad03211e",
"CoA-Custom-1.5-client-part2.zip" => "122f2882da255913e01610e33c6446ded897fecbe60b6d5ee026149becf8f278",
"CoA-Custom-1.6-main-20261012-4dd3a4ec.zip" => "dfe802ea70047c129ac2e0aaa557bab3110e3cc6926d36403b8c880a0e310ee4",
"CoA-Custom-1.6-client-part1.zip" => "1dbf1901ad5be61eba66973609df90712ffc78ef9e9b2c8549a5f3e36332bf12",
"CoA-Custom-1.6-client-part2.zip" => "689f7f946b4180fe977dabedc7cfe642e3e7f27431600c64c4dd5b7960b1bc66",
"CoA-Custom-1.7-client.zip" => "ad389ef486f44f269445f80807b25b214c39fc5d21272b5708bfeac8b1efefa0",
_ => return Err(Error::Invalid("Unknown custom races archive.".into())),
};
if fsx::sha256_file(path)? != expected {
@@ -660,7 +674,7 @@ mod tests {
}
#[test]
#[ignore = "Requires the three downloaded release archives"]
#[ignore = "Requires the downloaded release archives"]
fn bundled_hashes_match_verified_release_archives() {
use sha2::{Digest, Sha256};
let folder = PathBuf::from(std::env::var_os("COA_RACES_ARCHIVE_DIR").expect("COA_RACES_ARCHIVE_DIR"));
@@ -686,6 +700,33 @@ mod tests {
assert_eq!(actual, expected);
}
#[test]
fn legacy_installation_uses_its_own_hashes_when_preserving_originals() {
let temp = tempfile::tempdir().unwrap();
fs::write(temp.path().join(MANIFEST_FILENAME), "# CoA Custom 1.5\n").unwrap();
let legacy: std::collections::BTreeMap<String, String> = serde_json::from_str(include_str!("custom_races_hashes_1_5.json")).unwrap();
let current: std::collections::BTreeMap<String, String> = serde_json::from_str(include_str!("custom_races_hashes.json")).unwrap();
let files = patch_files(temp.path()).unwrap();
assert_eq!(files.iter().find(|file| file.path == "Ascension.exe").unwrap().sha256, legacy["Ascension.exe"]);
assert_eq!(files.len(), legacy.len());
assert!(current.len() > legacy.len());
assert!(!files.iter().any(|file| file.path == "Data/patch-ZZS.MPQ"));
}
#[test]
fn previous_release_uses_its_own_hashes_for_upgrade_and_removal() {
let temp = tempfile::tempdir().unwrap();
fs::write(temp.path().join(MANIFEST_FILENAME), "# CoA Custom 1.6\n").unwrap();
let legacy: std::collections::BTreeMap<String, String> = serde_json::from_str(include_str!("custom_races_hashes_1_6.json")).unwrap();
let current: std::collections::BTreeMap<String, String> = serde_json::from_str(include_str!("custom_races_hashes.json")).unwrap();
let files = patch_files(temp.path()).unwrap();
assert_eq!(files.len(), legacy.len());
assert_eq!(files.iter().find(|file| file.path == "Data/patch-T.MPQ").unwrap().sha256, legacy["Data/patch-T.MPQ"]);
assert_ne!(legacy["Data/patch-T.MPQ"], current["Data/patch-T.MPQ"]);
assert!(!files.iter().any(|file| file.path == "Data/patch-ZZW.MPQ"));
assert!(legacy.keys().all(|key| current.contains_key(key)));
}
#[test]
fn another_patch_operation_cannot_enter_the_same_client() {
let temp = tempfile::tempdir().unwrap();
@@ -855,4 +896,21 @@ mod tests {
assert!(client_path.join("dinput8.dll").is_file());
assert!(!client_path.join("dinput8.dll.disabled").is_file());
}
#[test]
fn disabling_bot_custom_races_excludes_every_nonstandard_race() {
let temp = tempfile::tempdir().unwrap();
let conf = temp.path().join("Core/configs/modules/playerbots.conf");
fs::create_dir_all(conf.parent().unwrap()).unwrap();
fs::write(&conf, "AiPlayerbot.ExcludedBotRaces = \"\"\n").unwrap();
sync_bot_races_config(temp.path(), false).unwrap();
assert_eq!(is_bot_custom_races_allowed(temp.path()), Some(false));
let parsed = crate::config::parser::ConfFile::parse_bytes(&fs::read(conf).unwrap()).unwrap();
let value = parsed.get("AiPlayerbot.ExcludedBotRaces").unwrap().trim_matches('"');
let excluded: Vec<u32> = value.split(',').map(|race| race.parse().unwrap()).collect();
for race in 1..=128 {
assert_eq!(excluded.contains(&race), ![1, 2, 3, 4, 5, 6, 7, 8, 10, 11].contains(&race), "race {race}");
}
}
}
+14 -9
View File
@@ -1,11 +1,4 @@
{
"Data/patch-ZE1.MPQ": "69d79dbbea54ec0312c37770debc79c7bc9ce2e21eedef99c03cd8d911955283",
"Data/patch-ZE2.MPQ": "1a41cae77ad730e9cf69e8d63fd12ae3a579418bb550d351fff469573a97ac3c",
"Data/patch-ZE3.MPQ": "b587202c21b6f580682b601b7c2e905fbfa766276e74733db55c1bddbffb7f87",
"Data/patch-ZEA.MPQ": "78fd10ec5262ae2417eba9700d2172553bbc1e5cfb0b597aada8f1753ca46d9d",
"Data/patch-ZHA.MPQ": "0fc6ebeafe1d14192dc64bafc90bde3c49230df1297f98bf65900a3bcf953617",
"Data/patch-ZZE.MPQ": "4b1e289e12c82644cb78f50170e6b5fd53918eb2de834f9e0cebddd81873a8bb",
"Data/patch-ZZR.MPQ": "5a7575e592dc0c8a6913e0d946bbe06fa0ca2e4d817ef1f8548d3d13142bcbd3",
"Ascension.exe": "e2c347019d554a59639e5978fd68e50e906973ab313c6573cc56ca82e8e3c547",
"EsteriaAppearance.bin": "b3fb7c69b7600c39f888ea1925701be5667097dcb02bb6ddd830e5489306bb50",
"EsteriaAppearance.dll": "392c5a2a25f9a69d705bdf841bb899aab768d4a3ee62734d576286c7198710d8",
@@ -34,7 +27,7 @@
"EsteriaVulpera.bin": "c5434993b9af4a17d881965e4735cd2ac379e39c7142aac9844adff9528562d4",
"EsteriaVulperaTextures.bin": "1ab844e74684aced2b069f1b91cd563aade45e05bd5e23f866a6c0c301cb098e",
"dinput8.dll": "f8de78606c241fb3911dce22faf68f5b9f9885b142b183c1e2c09ddc8f7e380c",
"Data/patch-T.MPQ": "9403e0680c74754eb0cb063718119aa6ace3616f7cecf488c8faf46f22505ae5",
"Data/patch-T.MPQ": "fe597138ab7f56881b5f4f27d6dc9d81bb278b483ca0d91b7207ffd4a5ae7bfd",
"Data/patch-ZE5.MPQ": "476b5a1fcf69f64fdb281e3b445362b936d016066ccd12f897acfcf200d6c0cf",
"Data/patch-ZE6.MPQ": "762592c9ed950b4a7b6c7d85d45050b30fa3f35d53e348ee1e1516fdbb7e5484",
"Data/patch-ZE7.MPQ": "de54b0189d242763d2872223419a259b6b8a4b380c24b50a2ef9c6fa44466248",
@@ -46,5 +39,17 @@
"Data/patch-ZHM.MPQ": "317d3ae00f0bd674b358af07544976acb7804aa6d8b19d1788ad96030e8c9bf0",
"Data/patch-ZJ.MPQ": "f821e99add86ef68d108c5ec466328c97677d3ed52f7944e200f3fe38c13b4de",
"Data/patch-ZM.MPQ": "424ab1978986e854551633983a94bd15beba6328aabea8c02f98eaa32f6f0b61",
"Data/patch-ZVU.MPQ": "c679b85a2e97344395756b54d2cf6abefe4e80e74944cb55b6e573ad17b7d020"
"Data/patch-ZVU.MPQ": "c679b85a2e97344395756b54d2cf6abefe4e80e74944cb55b6e573ad17b7d020",
"Data/patch-ZE1.MPQ": "69d79dbbea54ec0312c37770debc79c7bc9ce2e21eedef99c03cd8d911955283",
"Data/patch-ZE2.MPQ": "1a41cae77ad730e9cf69e8d63fd12ae3a579418bb550d351fff469573a97ac3c",
"Data/patch-ZE3.MPQ": "b587202c21b6f580682b601b7c2e905fbfa766276e74733db55c1bddbffb7f87",
"Data/patch-ZEA.MPQ": "78fd10ec5262ae2417eba9700d2172553bbc1e5cfb0b597aada8f1753ca46d9d",
"Data/patch-ZHA.MPQ": "0fc6ebeafe1d14192dc64bafc90bde3c49230df1297f98bf65900a3bcf953617",
"Data/patch-ZZE.MPQ": "4b1e289e12c82644cb78f50170e6b5fd53918eb2de834f9e0cebddd81873a8bb",
"Data/patch-ZZR.MPQ": "5a7575e592dc0c8a6913e0d946bbe06fa0ca2e4d817ef1f8548d3d13142bcbd3",
"Data/patch-ZZS.MPQ": "533fd69d5784d83e872a5a28d27301dc9ab5d0cb925f8b2116926854e4a0a5b8",
"Data/patch-ZZT.MPQ": "aef3bbe4951333d62fb55224484724fbaa9c35773a19944ffb9935e250cfb59c",
"Data/patch-ZZU.MPQ": "8ee1839415b56d46c6eff7eb4dfbaf6fbafc4c6b29b39380d69f4d827d02a99e",
"Data/patch-ZZV.MPQ": "8d5e61d37008abf70478e5d5f46ec7afb0490b9a643323bd75a1c70cadd63290",
"Data/patch-ZZW.MPQ": "ed452a0a0118c9f4deaf02db753064ac95b6e4a84022bdabd39a7e70d775223f"
}
@@ -0,0 +1,50 @@
{
"Data/patch-ZE1.MPQ": "69d79dbbea54ec0312c37770debc79c7bc9ce2e21eedef99c03cd8d911955283",
"Data/patch-ZE2.MPQ": "1a41cae77ad730e9cf69e8d63fd12ae3a579418bb550d351fff469573a97ac3c",
"Data/patch-ZE3.MPQ": "b587202c21b6f580682b601b7c2e905fbfa766276e74733db55c1bddbffb7f87",
"Data/patch-ZEA.MPQ": "78fd10ec5262ae2417eba9700d2172553bbc1e5cfb0b597aada8f1753ca46d9d",
"Data/patch-ZHA.MPQ": "0fc6ebeafe1d14192dc64bafc90bde3c49230df1297f98bf65900a3bcf953617",
"Data/patch-ZZE.MPQ": "4b1e289e12c82644cb78f50170e6b5fd53918eb2de834f9e0cebddd81873a8bb",
"Data/patch-ZZR.MPQ": "5a7575e592dc0c8a6913e0d946bbe06fa0ca2e4d817ef1f8548d3d13142bcbd3",
"Ascension.exe": "e2c347019d554a59639e5978fd68e50e906973ab313c6573cc56ca82e8e3c547",
"EsteriaAppearance.bin": "b3fb7c69b7600c39f888ea1925701be5667097dcb02bb6ddd830e5489306bb50",
"EsteriaAppearance.dll": "392c5a2a25f9a69d705bdf841bb899aab768d4a3ee62734d576286c7198710d8",
"EsteriaAppearanceGeometry.bin": "c95c53ca19c53c3434e12b99c1966f9c015ec3944c46dff7784d5e4ffc472d5d",
"EsteriaAppearanceMaterials.bin": "f72f027f54a55c05aa0f76b457b97b484fe0e1aedc0e068cc0e0ca53e1878cba",
"EsteriaCosmeticWings.bin": "57352c28459d073a29b3014d33b70f3b3756afa37fa9583ef6b30d503422ac3f",
"EsteriaEarthen.bin": "8a0db93ba2d5c756b44ed84f2f7bc7c15eef13be6a7d87baf1383c6c0c624e02",
"EsteriaGnoll.bin": "cfacc8e681bb8e54a4f01ee9c12a91bf0bd23a23527bb548d4dc21ec1b33eea2",
"EsteriaGnollTextures.bin": "1ba9133a75dad291d98a068b2dcde01f31a897a1764cd7f3e2c316c288a8f605",
"EsteriaHaranir.bin": "59f31d00e372fd2646e1cf0d9d3d5d671b31edd60ff7b862f8d1ce7e53ba1f2f",
"EsteriaHaranirTextures.bin": "99948ca570be2d376f0c67e3b8088739b6fa9f9d2c73985e19f68df77689a229",
"EsteriaHighmountain.bin": "e83349252ad2687ac23fcfc5623db7401cf32e5faf47a5a7be20c2f94b014938",
"EsteriaNaga.bin": "e87547cff5d09500c294f8b06935d134cdd0bbb962e294a72de09404505bd8b3",
"EsteriaNagaTextures.bin": "57ec560f99c1d48ac1e8f6a797c156513b99e7d349e3071afe405f812d159ae3",
"EsteriaSaberon.bin": "7735c9325c807c5125760d87f323d6b7d4162fb542c1d8b5f435ea2d6b9a4daa",
"EsteriaSaberonCoverage.bin": "9cdc227f08329e1245d2b864a996e7a772c96ecdbe5d81925f4d60cdbe4fa11e",
"EsteriaSaberonTextures.bin": "1ba9133a75dad291d98a068b2dcde01f31a897a1764cd7f3e2c316c288a8f605",
"EsteriaSethrak.bin": "bd14857a8c11fdf8027128763293701586b2102f9a6539d33a727014789fd606",
"EsteriaSethrakTextures.bin": "d2b2cf58087a391f00198008fe871edd400ddf28d54a3a05674df74c5198528a",
"EsteriaThinHuman.bin": "75794d192ea0c228b3da7bf7d0bdff1119f79d3f9bfef6650cd5e1a2b5d106eb",
"EsteriaThinHumanTextures.bin": "02aeb4c05893d06125b8b58fdb7c2a56fe3d66a7d094706a44b43cb503d1731e",
"EsteriaTuskarr.bin": "ecbe7ef490ba67eca029a57639192db61238522618491549766fe38529914387",
"EsteriaTuskarrTextures.bin": "e2116c2c83adea079126fd5da765d20da24f044d4f161c4124e4368cb3d6bc71",
"EsteriaVrykul.bin": "39e2b7b4c67997dbbd34bae840bd606a59290f7f2fc82565a0ba00090335415c",
"EsteriaVrykulTextures.bin": "46098750a0e235724c6e9cc8d9d4e26550aa47bef95fd5f0f82efe5e945bf296",
"EsteriaVulpera.bin": "c5434993b9af4a17d881965e4735cd2ac379e39c7142aac9844adff9528562d4",
"EsteriaVulperaTextures.bin": "1ab844e74684aced2b069f1b91cd563aade45e05bd5e23f866a6c0c301cb098e",
"dinput8.dll": "f8de78606c241fb3911dce22faf68f5b9f9885b142b183c1e2c09ddc8f7e380c",
"Data/patch-T.MPQ": "9403e0680c74754eb0cb063718119aa6ace3616f7cecf488c8faf46f22505ae5",
"Data/patch-ZE5.MPQ": "476b5a1fcf69f64fdb281e3b445362b936d016066ccd12f897acfcf200d6c0cf",
"Data/patch-ZE6.MPQ": "762592c9ed950b4a7b6c7d85d45050b30fa3f35d53e348ee1e1516fdbb7e5484",
"Data/patch-ZE7.MPQ": "de54b0189d242763d2872223419a259b6b8a4b380c24b50a2ef9c6fa44466248",
"Data/patch-ZE8.MPQ": "1bc5691fb84a84f3e7a0a70baaa9192a97b4ec5906ee37c85838ffe08adef934",
"Data/patch-ZE9.MPQ": "d49c94463e9548e76378ce2c0454d8f204f0745c8c0de32b6dce7f0b0fd67c4a",
"Data/patch-ZES.MPQ": "a2d9348f917f745c7aa25b2eed96efa312f7b2d2f62cb86a4f2eaabc2dfa895f",
"Data/patch-ZG.MPQ": "48979e1fda775f8deb47bffa48e0a468595d4c71286fee518f51dc5801c5ca39",
"Data/patch-ZH.MPQ": "5e1d2fe287053abba703d7af381e10096c6525ba477ffed3971c2fc4f93e919c",
"Data/patch-ZHM.MPQ": "317d3ae00f0bd674b358af07544976acb7804aa6d8b19d1788ad96030e8c9bf0",
"Data/patch-ZJ.MPQ": "f821e99add86ef68d108c5ec466328c97677d3ed52f7944e200f3fe38c13b4de",
"Data/patch-ZM.MPQ": "424ab1978986e854551633983a94bd15beba6328aabea8c02f98eaa32f6f0b61",
"Data/patch-ZVU.MPQ": "c679b85a2e97344395756b54d2cf6abefe4e80e74944cb55b6e573ad17b7d020"
}
@@ -0,0 +1,52 @@
{
"Ascension.exe": "e2c347019d554a59639e5978fd68e50e906973ab313c6573cc56ca82e8e3c547",
"EsteriaAppearance.bin": "b3fb7c69b7600c39f888ea1925701be5667097dcb02bb6ddd830e5489306bb50",
"EsteriaAppearance.dll": "392c5a2a25f9a69d705bdf841bb899aab768d4a3ee62734d576286c7198710d8",
"EsteriaAppearanceGeometry.bin": "c95c53ca19c53c3434e12b99c1966f9c015ec3944c46dff7784d5e4ffc472d5d",
"EsteriaAppearanceMaterials.bin": "f72f027f54a55c05aa0f76b457b97b484fe0e1aedc0e068cc0e0ca53e1878cba",
"EsteriaCosmeticWings.bin": "57352c28459d073a29b3014d33b70f3b3756afa37fa9583ef6b30d503422ac3f",
"EsteriaEarthen.bin": "8a0db93ba2d5c756b44ed84f2f7bc7c15eef13be6a7d87baf1383c6c0c624e02",
"EsteriaGnoll.bin": "cfacc8e681bb8e54a4f01ee9c12a91bf0bd23a23527bb548d4dc21ec1b33eea2",
"EsteriaGnollTextures.bin": "1ba9133a75dad291d98a068b2dcde01f31a897a1764cd7f3e2c316c288a8f605",
"EsteriaHaranir.bin": "59f31d00e372fd2646e1cf0d9d3d5d671b31edd60ff7b862f8d1ce7e53ba1f2f",
"EsteriaHaranirTextures.bin": "99948ca570be2d376f0c67e3b8088739b6fa9f9d2c73985e19f68df77689a229",
"EsteriaHighmountain.bin": "e83349252ad2687ac23fcfc5623db7401cf32e5faf47a5a7be20c2f94b014938",
"EsteriaNaga.bin": "e87547cff5d09500c294f8b06935d134cdd0bbb962e294a72de09404505bd8b3",
"EsteriaNagaTextures.bin": "57ec560f99c1d48ac1e8f6a797c156513b99e7d349e3071afe405f812d159ae3",
"EsteriaSaberon.bin": "7735c9325c807c5125760d87f323d6b7d4162fb542c1d8b5f435ea2d6b9a4daa",
"EsteriaSaberonCoverage.bin": "9cdc227f08329e1245d2b864a996e7a772c96ecdbe5d81925f4d60cdbe4fa11e",
"EsteriaSaberonTextures.bin": "1ba9133a75dad291d98a068b2dcde01f31a897a1764cd7f3e2c316c288a8f605",
"EsteriaSethrak.bin": "bd14857a8c11fdf8027128763293701586b2102f9a6539d33a727014789fd606",
"EsteriaSethrakTextures.bin": "d2b2cf58087a391f00198008fe871edd400ddf28d54a3a05674df74c5198528a",
"EsteriaThinHuman.bin": "75794d192ea0c228b3da7bf7d0bdff1119f79d3f9bfef6650cd5e1a2b5d106eb",
"EsteriaThinHumanTextures.bin": "02aeb4c05893d06125b8b58fdb7c2a56fe3d66a7d094706a44b43cb503d1731e",
"EsteriaTuskarr.bin": "ecbe7ef490ba67eca029a57639192db61238522618491549766fe38529914387",
"EsteriaTuskarrTextures.bin": "e2116c2c83adea079126fd5da765d20da24f044d4f161c4124e4368cb3d6bc71",
"EsteriaVrykul.bin": "39e2b7b4c67997dbbd34bae840bd606a59290f7f2fc82565a0ba00090335415c",
"EsteriaVrykulTextures.bin": "46098750a0e235724c6e9cc8d9d4e26550aa47bef95fd5f0f82efe5e945bf296",
"EsteriaVulpera.bin": "c5434993b9af4a17d881965e4735cd2ac379e39c7142aac9844adff9528562d4",
"EsteriaVulperaTextures.bin": "1ab844e74684aced2b069f1b91cd563aade45e05bd5e23f866a6c0c301cb098e",
"dinput8.dll": "f8de78606c241fb3911dce22faf68f5b9f9885b142b183c1e2c09ddc8f7e380c",
"Data/patch-T.MPQ": "9403e0680c74754eb0cb063718119aa6ace3616f7cecf488c8faf46f22505ae5",
"Data/patch-ZE5.MPQ": "476b5a1fcf69f64fdb281e3b445362b936d016066ccd12f897acfcf200d6c0cf",
"Data/patch-ZE6.MPQ": "762592c9ed950b4a7b6c7d85d45050b30fa3f35d53e348ee1e1516fdbb7e5484",
"Data/patch-ZE7.MPQ": "de54b0189d242763d2872223419a259b6b8a4b380c24b50a2ef9c6fa44466248",
"Data/patch-ZE8.MPQ": "1bc5691fb84a84f3e7a0a70baaa9192a97b4ec5906ee37c85838ffe08adef934",
"Data/patch-ZE9.MPQ": "d49c94463e9548e76378ce2c0454d8f204f0745c8c0de32b6dce7f0b0fd67c4a",
"Data/patch-ZES.MPQ": "a2d9348f917f745c7aa25b2eed96efa312f7b2d2f62cb86a4f2eaabc2dfa895f",
"Data/patch-ZG.MPQ": "48979e1fda775f8deb47bffa48e0a468595d4c71286fee518f51dc5801c5ca39",
"Data/patch-ZH.MPQ": "5e1d2fe287053abba703d7af381e10096c6525ba477ffed3971c2fc4f93e919c",
"Data/patch-ZHM.MPQ": "317d3ae00f0bd674b358af07544976acb7804aa6d8b19d1788ad96030e8c9bf0",
"Data/patch-ZJ.MPQ": "f821e99add86ef68d108c5ec466328c97677d3ed52f7944e200f3fe38c13b4de",
"Data/patch-ZM.MPQ": "424ab1978986e854551633983a94bd15beba6328aabea8c02f98eaa32f6f0b61",
"Data/patch-ZVU.MPQ": "c679b85a2e97344395756b54d2cf6abefe4e80e74944cb55b6e573ad17b7d020",
"Data/patch-ZE1.MPQ": "69d79dbbea54ec0312c37770debc79c7bc9ce2e21eedef99c03cd8d911955283",
"Data/patch-ZE2.MPQ": "1a41cae77ad730e9cf69e8d63fd12ae3a579418bb550d351fff469573a97ac3c",
"Data/patch-ZE3.MPQ": "b587202c21b6f580682b601b7c2e905fbfa766276e74733db55c1bddbffb7f87",
"Data/patch-ZEA.MPQ": "78fd10ec5262ae2417eba9700d2172553bbc1e5cfb0b597aada8f1753ca46d9d",
"Data/patch-ZHA.MPQ": "0fc6ebeafe1d14192dc64bafc90bde3c49230df1297f98bf65900a3bcf953617",
"Data/patch-ZZE.MPQ": "4b1e289e12c82644cb78f50170e6b5fd53918eb2de834f9e0cebddd81873a8bb",
"Data/patch-ZZR.MPQ": "5a7575e592dc0c8a6913e0d946bbe06fa0ca2e4d817ef1f8548d3d13142bcbd3",
"Data/patch-ZZS.MPQ": "533fd69d5784d83e872a5a28d27301dc9ab5d0cb925f8b2116926854e4a0a5b8",
"Data/patch-ZZT.MPQ": "aef3bbe4951333d62fb55224484724fbaa9c35773a19944ffb9935e250cfb59c"
}
+54 -1
View File
@@ -109,7 +109,8 @@ pub fn run(root: &Path, meta: &InstallMeta) -> Report {
c.push(if crate::client::detect(Path::new(path), None).is_some() { check("client", "Game client", Level::Ok, path.clone()) } else { check("client", "Game client", Level::Warn, "The saved game folder was not found; choose it again in Settings.") });
}
let exposed: Vec<&str> = crate::net::exposure(&ports).iter().filter(|e| (e.what == "database" || e.what == "server console") && e.reachable_from_network).map(|e| e.what).collect();
let exposure = if crate::docker::is_docker(root) { crate::docker::exposure(root, &ports) } else { crate::net::exposure(&ports) };
let exposed: Vec<&str> = exposure.iter().filter(|e| (e.what == "database" || e.what == "server console") && e.reachable_from_network).map(|e| e.what).collect();
c.push(if exposed.is_empty() { check("exposure", "Private services", Level::Ok, "Database and server console are not reachable from the network.") } else { check("exposure", "Private services", Level::Fail, format!("Reachable from the network: {}.", exposed.join(", "))) });
let problems = c.iter().filter(|x| x.level != Level::Ok).count();
@@ -347,6 +348,15 @@ pub fn export_package(root: &Path, meta_dir: &Path, manager_log: &Path, meta: &I
"wow_client_set": meta.client_path.is_some(),
});
add("summary.json", serde_json::to_string_pretty(&summary)?.as_bytes())?;
let owner = fs::read_to_string(meta_dir.join("update-lock-owner.json")).ok()
.and_then(|text| serde_json::from_str::<serde_json::Value>(&redact_json(&text)).ok());
let maintenance = serde_json::json!({
"lockFilePresent": meta_dir.join("update.lock").is_file(),
"lastRecordedOwner": owner,
"note": "The lock file and owner record may remain after completion. Their presence does not establish whether the OS lock is currently held."
});
add("maintenance.json", serde_json::to_string_pretty(&maintenance)?.as_bytes())?;
// Logs. The newest update logs of the repack's own updater are added under their own names.
let mut logs: Vec<(String, PathBuf)> = vec![
@@ -464,16 +474,53 @@ pub fn stamp() -> String {
chrono::Local::now().format("%Y%m%d-%H%M%S").to_string()
}
#[cfg(windows)]
pub fn desktop_or_temp() -> PathBuf {
let d = std::env::var_os("USERPROFILE").map(PathBuf::from).unwrap_or_else(std::env::temp_dir).join("Desktop");
if d.is_dir() { d } else { std::env::temp_dir() }
}
/// Where a file meant for the person goes on Linux: the desktop folder they have (the XDG user directories file names it), else
/// `~/Desktop`, else `~/Downloads`, else the temporary folder.
#[cfg(not(windows))]
pub fn desktop_or_temp() -> PathBuf {
let Some(home) = std::env::var_os("HOME").map(PathBuf::from) else { return std::env::temp_dir() };
let dirs = std::fs::read_to_string(home.join(".config/user-dirs.dirs")).ok();
pick_desktop(&home, dirs.as_deref(), &|p| p.is_dir())
}
#[cfg(not(windows))]
fn pick_desktop(home: &Path, user_dirs: Option<&str>, exists: &dyn Fn(&Path) -> bool) -> PathBuf {
let named = user_dirs.and_then(|t| {
t.lines().find_map(|l| {
let v = l.trim().strip_prefix("XDG_DESKTOP_DIR=")?.trim_matches('"');
Some(PathBuf::from(v.replace("$HOME", &home.to_string_lossy())))
})
});
// A user without a desktop sets XDG_DESKTOP_DIR to the home folder itself: that is not a place to drop files into.
[named, Some(home.join("Desktop")), Some(home.join("Downloads"))].into_iter().flatten().filter(|p| p != home).find(|p| exists(p)).unwrap_or_else(std::env::temp_dir)
}
#[cfg(test)]
mod tests {
use super::*;
use crate::registry::{InstallKind, InstallMeta};
#[cfg(not(windows))]
#[test]
fn the_desktop_folder_on_linux_follows_the_users_own_directories() {
let home = Path::new("/home/ana");
let only = |wanted: &'static [&'static str]| move |p: &Path| wanted.contains(&p.to_str().unwrap());
let dirs = "# comment\nXDG_DOWNLOAD_DIR=\"$HOME/Downloads\"\nXDG_DESKTOP_DIR=\"$HOME/Bureau\"\n";
assert_eq!(pick_desktop(home, Some(dirs), &only(&["/home/ana/Bureau", "/home/ana/Desktop"])), PathBuf::from("/home/ana/Bureau"));
assert_eq!(pick_desktop(home, None, &only(&["/home/ana/Desktop"])), PathBuf::from("/home/ana/Desktop"));
assert_eq!(pick_desktop(home, Some(dirs), &only(&["/home/ana/Downloads"])), PathBuf::from("/home/ana/Downloads"));
assert_eq!(pick_desktop(home, None, &only(&[])), std::env::temp_dir());
// No desktop: the user directories file names the home folder itself.
let none = "XDG_DESKTOP_DIR=\"$HOME/\"\n";
assert_eq!(pick_desktop(home, Some(none), &only(&["/home/ana", "/home/ana/Downloads"])), PathBuf::from("/home/ana/Downloads"));
}
#[test]
fn repeated_config_warnings_are_folded() {
let t = "start
@@ -599,6 +646,8 @@ real error
fs::write(root.join("Core/configs/worldserver.conf"), "LoginDatabaseInfo = \"127.0.0.1;3307;acore;SECRETPW;auth\"\nRate.XP.Kill = 1\n").unwrap();
let meta_dir = d.path().join("srv.manager");
fs::create_dir_all(&meta_dir).unwrap();
fs::write(meta_dir.join("update.lock"), b"").unwrap();
fs::write(meta_dir.join("update-lock-owner.json"), r#"{"operation":"checking updates","pid":1234,"started":"2026-10-13T01:00:00Z","token":"owner-secret"}"#).unwrap();
fs::create_dir_all(meta_dir.join("diagnostics")).unwrap();
fs::write(meta_dir.join("diagnostics/schema-validation-CoA-test.json"), r#"{"problems":[{"table":"creature_display_preset","column":"display_id","detail":"default mismatch"}],"password":"schema-secret"}"#).unwrap();
fs::write(meta_dir.join("diagnostics/database-recovery-test.json"), r#"{"objects":[{"database":"acore_characters","kind":"TRIGGER","name":"user_character_trigger"}],"password":"recovery-secret"}"#).unwrap();
@@ -620,6 +669,10 @@ real error
assert!(!all.contains("schema-secret"));
assert!(all.contains("user_character_trigger") && all.contains("acore_characters"));
assert!(!all.contains("recovery-secret"));
assert!(all.contains("checking updates") && all.contains("1234"));
assert!(all.contains("lockFilePresent") && all.contains("does not establish"));
assert!(!all.contains("owner-secret"));
assert_eq!(fs::read(meta_dir.join("update.lock")).unwrap(), b"");
assert!(!all.contains("hunter2") && !all.contains("SECRETPW") && !all.contains("upstream-secret"));
}
}
+1 -1
View File
@@ -112,7 +112,7 @@ pub fn preflight(dest: &Path, data_dir: &Path, needed_bytes: u64, registry: &Reg
Preflight { ok: problems.is_empty(), problems, free_bytes }
}
fn data_problems(data_dir: &Path) -> Vec<crate::install::Problem> {
pub(crate) fn data_problems(data_dir: &Path) -> Vec<crate::install::Problem> {
let s = data_dir.to_string_lossy();
if !data_dir.is_absolute() || s.contains(':') {
return vec![problem("data_path", "Choose the game data folder with its full path, without a colon in it.")];
+57
View File
@@ -242,6 +242,10 @@ fn start(d: &dyn Docker, root: &Path, cfg: &Config, with_game: bool, log: &mut L
run_container(d, &call, &n.world, log)?;
}
wait_listening(d, cfg, &n.world, ports.world, WORLD_WAIT, log)?;
// The core leaves with code 2 for `server restart` and expects a supervisor to start it again (the repack has one). Docker
// does that for any non-zero exit, but only once the server has come up: a start that fails must stay failed, not loop.
// A stop by `docker stop` or `server shutdown` (code 0) is never restarted.
let _ = docker(d, &["update", "--restart", "on-failure:3", &n.world], QUICK);
if !state.get(&n.auth).is_some_and(Container::running) {
remove(d, &n.auth);
@@ -251,9 +255,26 @@ fn start(d: &dyn Docker, root: &Path, cfg: &Config, with_game: bool, log: &mut L
run_container(d, &call, &n.auth, log)?;
}
wait_listening(d, cfg, &n.auth, ports.auth, AUTH_WAIT, log)?;
mark_realm_online(d, &n, &secrets, realm, log);
Ok(())
}
/// The auth server marks every realm offline when it starts, and the world server (already up, see the order above) only
/// puts itself back after its next uptime update, ten minutes later by default: until then the client shows "Realm is
/// Offline". The world is known to be listening here, so say so for the realm it serves. The other realm of a Wildcard
/// install keeps the offline flag it is given on purpose (`realms::setup_realmlist`).
fn mark_realm_online(d: &dyn Docker, n: &Names, secrets: &Secrets, realm: crate::realms::Mode, log: &mut Log) {
let sql = format!("UPDATE acore_auth.realmlist SET flag = flag & ~2 WHERE id = {};", realm.realm_id());
let mut call = Call::new(&["exec", "-i", "-e", "MYSQL_PWD", &n.db, "mysql", "--user=root", "--protocol=tcp", "--host=127.0.0.1"], QUICK);
call.env = vec![("MYSQL_PWD".into(), secrets.root.clone())];
call.stdin = Some(sql.as_bytes());
match d.run(&call) {
Ok(o) if o.ok() => {}
Ok(o) => log.say(format!("the realm could not be marked online: {}", o.text().trim())),
Err(e) => log.say(format!("the realm could not be marked online: {e}")),
}
}
fn remove(d: &dyn Docker, name: &str) {
// A stopped container of the same name is replaced, so its settings always match the files on disk.
let _ = d.run(&Call::new(&["rm", "-f", name], QUICK));
@@ -388,6 +409,21 @@ fn database_env(s: &Secrets, world: bool, realm: crate::realms::Mode) -> Vec<(St
env
}
/// Every setting variable (`AC_...`) the containers get, for the check that the screens know which settings are fixed.
#[cfg(test)]
pub(crate) fn forced_variables() -> Vec<String> {
let cfg = Config { project: "t".into(), bind_address: "127.0.0.1".into(), mysql_image: "mysql:8.4".into(), data_dir: None, mysql_data: None };
let n = cfg.names();
let mut found = Vec::new();
for kind in [GameKind::World, GameKind::Auth] {
let args = game_args(&cfg, &n, kind, Path::new("/s"), Path::new("/d"), 1, 2, None);
found.extend(args.iter().filter_map(|a| a.strip_prefix("AC_").map(|rest| format!("AC_{}", rest.split('=').next().unwrap()))));
}
found.sort();
found.dedup();
found
}
/// Realm profiles are only in play once a profile has been chosen on this server (`Settings/realm-profile.json`).
fn prepare_realm(root: &Path, log: &mut Log) -> Step<crate::realms::Mode> {
let fail_with = |e: crate::error::Error| Failure { code: ErrorCode::StartupFailed, output: e.to_string() };
@@ -670,6 +706,8 @@ mod tests {
}
out(0, "containerid\n", "")
}
["update", "--restart", _, _] => out(0, "", ""),
["exec", ..] => out(0, "", ""),
["stop", "--time", _, name] => {
self.running.borrow_mut().remove(*name);
self.exited.borrow_mut().insert(name.to_string(), (0, false));
@@ -712,6 +750,10 @@ mod tests {
assert!(at("network create") < at("run --detach"), "database comes first");
let runs: Vec<String> = sim.calls_of("run").iter().map(|c| c.args[c.args.iter().position(|x| x == "--name").unwrap() + 1].clone()).collect();
assert_eq!(runs, ["coa-t1-db", "coa-t1-world", "coa-t1-auth"], "database, then world, then auth");
let exec = sim.calls_of("exec");
assert_eq!(exec.len(), 1, "the realm is marked online once the auth server is up");
assert!(exec[0].stdin && exec[0].env == [("MYSQL_PWD".to_string(), ROOT_PW.to_string())]);
assert!(!exec[0].args.iter().any(|a| a.contains(ROOT_PW)), "the password is not on the command line");
assert!(at("build") < verbs.iter().rposition(|v| v.starts_with("run")).unwrap(), "the runtime image exists before the game servers");
assert!(sim.calls_of("build")[0].stdin, "the Dockerfile travels on stdin");
}
@@ -757,6 +799,21 @@ mod tests {
assert!(!plain.iter().any(|a| a == "--lower-case-table-names=1" || a == "--user"));
}
#[test]
fn the_world_server_is_started_again_when_it_restarts_itself_but_only_once_it_is_up() {
let (_d, root) = server("t1");
let sim = Sim::new();
assert!(start_all(&sim, &root).ok);
let world_run = sim.calls_of("run").into_iter().find(|c| c.args.iter().any(|a| a == "coa-t1-world")).unwrap();
assert!(!world_run.args.iter().any(|a| a == "--restart"), "no restart while it is still starting");
let updates = sim.calls_of("update");
assert_eq!(updates.len(), 1);
assert_eq!(updates[0].args, ["update", "--restart", "on-failure:3", "coa-t1-world"]);
let verbs = sim.verbs();
let pos = |p: &str| verbs.iter().position(|v| v.starts_with(p)).unwrap();
assert!(pos("run --detach") < pos("update"), "after the container exists and listens");
}
#[test]
fn the_containers_use_the_databases_of_the_selected_realm() {
let s = Secrets { root: "r".into(), app: "apppw".into() };
+72
View File
@@ -0,0 +1,72 @@
//! The database of a Docker server has no log file in the server folder: MySQL writes to its container's output.
use super::cli::{Call, Docker, SystemDocker};
use super::Config;
use crate::console::Line;
use crate::error::{Error, Result};
use std::path::Path;
use std::time::Duration;
/// The last lines MySQL wrote (what `docker logs` returns), filtered and classified like the lines of a log file.
pub fn database_log(root: &Path, filter: Option<&str>, max_lines: usize) -> Result<Vec<Line>> {
database_log_with(&SystemDocker, root, filter, max_lines)
}
pub fn database_log_with(d: &dyn Docker, root: &Path, filter: Option<&str>, max_lines: usize) -> Result<Vec<Line>> {
let name = Config::load(root)?.names().db;
// Ask for more than wanted: the filter may drop most of them.
let tail = (max_lines * 4).clamp(200, 8000).to_string();
let out = d.run(&Call::new(&["logs", "--tail", &tail, &name], Duration::from_secs(20)))?;
if !out.ok() {
return Err(Error::Invalid("This log does not exist yet.".into()));
}
// MySQL writes to the error stream.
Ok(crate::console::lines_from_text(&format!("{}\n{}", out.stderr, out.stdout), filter, max_lines))
}
#[cfg(test)]
mod tests {
use super::*;
use crate::docker::cli::Output;
use std::cell::RefCell;
struct Fake {
out: Output,
calls: RefCell<Vec<Vec<String>>>,
}
impl Docker for Fake {
fn run(&self, call: &Call) -> Result<Output> {
self.calls.borrow_mut().push(call.args.clone());
Ok(self.out.clone())
}
}
fn root() -> (tempfile::TempDir, std::path::PathBuf) {
let d = tempfile::tempdir().unwrap();
let root = d.path().join("srv");
std::fs::create_dir_all(root.join("Settings")).unwrap();
std::fs::write(root.join("Settings/docker.json"), r#"{"project":"t1"}"#).unwrap();
(d, root)
}
#[test]
fn the_database_log_is_what_the_container_printed() {
let (_d, root) = root();
let fake = Fake { out: Output { code: Some(0), stdout: "2026-10-04T10:00:00Z [System] ready\n".into(), stderr: "2026-10-04T10:00:01Z [ERROR] [MY-010] boom\n2026-10-04T10:00:02Z [Warning] slow\n".into() }, calls: Default::default() };
let lines = database_log_with(&fake, &root, None, 100).unwrap();
assert_eq!(fake.calls.borrow()[0][..2], ["logs", "--tail"]);
assert_eq!(fake.calls.borrow()[0].last().unwrap(), "coa-t1-db");
assert_eq!(lines.len(), 3);
assert!(lines.iter().any(|l| l.text.contains("boom") && l.level == crate::console::Level::Error));
let only = database_log_with(&fake, &root, Some("slow"), 100).unwrap();
assert_eq!(only.len(), 1);
}
#[test]
fn a_database_that_never_ran_has_no_log_yet() {
let (_d, root) = root();
let fake = Fake { out: Output { code: Some(1), stdout: String::new(), stderr: "Error: No such container: coa-t1-db".into() }, calls: Default::default() };
assert!(database_log_with(&fake, &root, None, 10).unwrap_err().to_string().contains("does not exist yet"));
}
}
+141
View File
@@ -18,9 +18,12 @@ mod cli;
pub mod fixture;
pub mod install;
mod lifecycle;
pub mod logs;
pub mod sharing;
pub use cli::{Call, Docker, Output, SystemDocker};
pub use lifecycle::{check_docker, observe, observe_with, run, run_with};
pub use sharing::{bind_is_open, exposure, set_open};
pub(crate) use lifecycle::destroy;
use std::net::IpAddr;
@@ -140,6 +143,84 @@ pub fn ensure_main_configs(root: &Path) -> Result<Vec<String>> {
Ok(created)
}
/// Settings the Docker backend gives the containers as environment variables. The core reads `AC_<KEY>` before the
/// configuration file, so a value written to the file for one of these keys is silently ignored. They are what makes the
/// containers work (where things are inside them, listening on every interface, the console, no built-in database
/// updater, the databases); the owner has no use for them and the screens do not offer them.
pub(crate) const FORCED_KEYS: [&str; 12] = [
"BindIP",
"WorldServerPort",
"RealmServerPort",
"LogsDir",
"DataDir",
"Updates.EnableDatabases",
"Ra.Enable",
"Ra.IP",
"Ra.Port",
"LoginDatabaseInfo",
"WorldDatabaseInfo",
"CharacterDatabaseInfo",
];
/// Is this setting fixed by the Docker backend (see `FORCED_KEYS`)?
pub fn managed_setting(key: &str) -> bool {
FORCED_KEYS.iter().any(|k| k.eq_ignore_ascii_case(key))
}
/// The environment variable the core reads for a setting: `AC_` and the key in upper snake case, a `_` between a lower-case
/// letter and a capital, and at letter / digit boundaries (the core's `IniKeyToEnvVarKey`).
#[cfg(test)]
pub(crate) fn env_name(key: &str) -> String {
let chars: Vec<char> = key.chars().collect();
let mut out = String::from("AC_");
for (i, &c) in chars.iter().enumerate() {
if matches!(c, ' ' | '.' | '-') {
out.push('_');
continue;
}
out.push(c.to_ascii_uppercase());
if let Some(&next) = chars.get(i + 1) {
let split = (!c.is_ascii_uppercase() && next.is_ascii_uppercase()) || (!c.is_ascii_digit() && next.is_ascii_digit()) || (c.is_ascii_digit() && !next.is_ascii_digit() && !matches!(next, ' ' | '.' | '-'));
if split {
out.push('_');
}
}
}
out
}
/// The folder mounted as the game data of this Docker server.
pub fn game_data_dir(root: &Path) -> Result<std::path::PathBuf> {
Ok(Config::load(root)?.data_path(root))
}
/// Point a Docker server at another game data folder (used in place, read-only). The containers are recreated at the next
/// start, so the servers must be stopped: a running one keeps the folder it has.
pub fn set_game_data(root: &Path, path: &Path) -> Result<()> {
let problems = install::data_problems(path);
if !problems.is_empty() {
return Err(Error::Invalid(problems.iter().map(|p| p.message.clone()).collect::<Vec<_>>().join(" ")));
}
let observed = crate::process::observe(root, &crate::layout::read_ports(root));
let busy = |s: &crate::process::ServiceStatus| matches!(s.state, crate::process::ServiceState::Running | crate::process::ServiceState::Starting | crate::process::ServiceState::Stopping);
if busy(&observed.world) || busy(&observed.auth) {
return Err(Error::Invalid("Stop the server before changing the game data folder.".into()));
}
let mut cfg = Config::load(root)?;
cfg.data_dir = Some(path.to_string_lossy().into_owned());
cfg.validate()?;
fsx::atomic_write_json(&root.join(MARKER), &cfg)
}
/// A reason a value cannot work on a Docker server, for settings that name a path: the server runs in a container that sees only
/// its own `Core` folder (as `/srv/core`, its working directory), so a path elsewhere on this computer does not exist for it.
pub fn setting_problem(key: &str, raw: &str) -> Option<&'static str> {
let raw = raw.trim().trim_matches('"');
let outside = Path::new(raw).is_absolute() && !raw.starts_with("/srv/core/");
(key.eq_ignore_ascii_case("CoaBots.TalentBuildsPath") && !raw.is_empty() && outside)
.then_some("must be a path inside the server's Core folder (for example reference/ascensionsidekick-level-builds.json), because the server runs in a container that cannot see other folders")
}
/// Name of the runtime image: it follows the content of its Dockerfile, so changing the libraries builds a new image.
pub(crate) fn runtime_image() -> String {
format!("coa-runtime:{}", &fsx::sha256_bytes(RUNTIME_DOCKERFILE.as_bytes())[..12])
@@ -206,6 +287,66 @@ mod tests {
use super::*;
use std::fs;
#[cfg(unix)]
#[test]
fn the_game_data_folder_can_be_changed_when_the_servers_are_stopped() {
let d = tempfile::tempdir().unwrap();
let root = d.path().join("srv");
std::fs::create_dir_all(root.join("Settings")).unwrap();
std::fs::write(root.join(MARKER), r#"{"project":"t1","dataDir":"/old/data"}"#).unwrap();
let good = d.path().join("data");
std::fs::create_dir_all(good.join("dbc")).unwrap();
std::fs::create_dir_all(good.join("maps")).unwrap();
assert_eq!(game_data_dir(&root).unwrap(), std::path::PathBuf::from("/old/data"));
set_game_data(&root, &good).unwrap();
assert_eq!(game_data_dir(&root).unwrap(), good);
assert_eq!(Config::load(&root).unwrap().project, "t1", "the other settings are kept");
// A folder that does not look like game data is refused and nothing changes.
let empty = d.path().join("empty");
std::fs::create_dir_all(&empty).unwrap();
assert!(set_game_data(&root, &empty).is_err());
assert_eq!(game_data_dir(&root).unwrap(), good);
assert!(set_game_data(&root, std::path::Path::new("relative/dir")).is_err());
}
#[cfg(unix)]
#[test]
fn a_path_outside_the_core_folder_is_refused_for_a_docker_server() {
let key = "CoaBots.TalentBuildsPath";
assert!(setting_problem(key, "reference/builds.json").is_none());
assert!(setting_problem(key, "").is_none());
assert!(setting_problem(key, "\"/srv/core/reference/builds.json\"").is_none());
assert!(setting_problem(key, "/home/ana/builds.json").is_some());
assert!(setting_problem("Rate.XP.Kill", "/home/ana/x").is_none(), "only settings that name a path");
}
#[test]
fn setting_names_become_the_variables_the_core_reads() {
for (key, var) in [
("BindIP", "AC_BIND_IP"),
("Ra.Enable", "AC_RA_ENABLE"),
("Ra.IP", "AC_RA_IP"),
("Updates.EnableDatabases", "AC_UPDATES_ENABLE_DATABASES"),
("WorldServerPort", "AC_WORLD_SERVER_PORT"),
("LoginDatabaseInfo", "AC_LOGIN_DATABASE_INFO"),
("DataDir", "AC_DATA_DIR"),
("Dynamic.XP.Reminder.Interval", "AC_DYNAMIC_XP_REMINDER_INTERVAL"),
("EtherealBazaar.Enable", "AC_ETHEREAL_BAZAAR_ENABLE"),
] {
assert_eq!(env_name(key), var, "{key}");
}
}
#[test]
fn the_settings_called_managed_are_exactly_the_variables_the_containers_get() {
// If a variable is added to the containers without being listed here (or the other way round), a screen would
// offer a setting that does nothing, or hide one that works.
let given: std::collections::BTreeSet<String> = lifecycle::forced_variables().into_iter().collect();
let listed: std::collections::BTreeSet<String> = FORCED_KEYS.iter().map(|k| env_name(k)).collect();
assert_eq!(given, listed);
assert!(managed_setting("ra.enable") && managed_setting("Updates.EnableDatabases") && !managed_setting("PlayerLimit"));
}
#[test]
fn the_main_configs_are_created_from_their_templates_and_never_overwritten() {
let d = tempfile::tempdir().unwrap();
+149
View File
@@ -0,0 +1,149 @@
//! Sharing a Docker server with friends. Who can reach the game is decided by the address the container ports are
//! published on (`bindAddress` in `Settings/docker.json`): `127.0.0.1` is this computer only, anything else is the network.
//! The `BindIP` of the configuration files is forced to `0.0.0.0` inside the containers and changes nothing, so the
//! repack's way of opening the servers (editing `BindIP`) is replaced here. The database is never published and the server
//! console is only ever published on the loopback address.
use std::fs;
use std::net::IpAddr;
use std::path::{Path, PathBuf};
use super::{Config, MARKER};
use crate::config::parser::ConfFile;
use crate::config::{take_snapshot, Scope};
use crate::error::Result;
use crate::fsx;
use crate::layout::Ports;
use crate::net::Exposure;
use crate::process::{ServiceState, ServiceStatus};
fn is_open(cfg: &Config) -> bool {
cfg.bind_address.parse::<IpAddr>().map(|ip| !ip.is_loopback()).unwrap_or(false)
}
/// Whether the login and world servers are published beyond this computer.
pub fn bind_is_open(root: &Path) -> bool {
Config::load(root).map(|c| is_open(&c)).unwrap_or(false)
}
/// Publish the login and world servers beyond this computer (`0.0.0.0`) or only on it (`127.0.0.1`). An address the owner chose
/// by hand (for example one network card) is kept when opening. Returns true if anything changed, which needs a restart of
/// the containers. The old files are snapshotted.
pub fn set_open(root: &Path, meta: &Path, open: bool) -> Result<bool> {
let mut cfg = Config::load(root)?;
let mut originals: Vec<(PathBuf, Vec<u8>)> = Vec::new();
let mut changed = false;
if is_open(&cfg) != open {
originals.push((root.join(MARKER), fs::read(root.join(MARKER))?));
cfg.bind_address = if open { "0.0.0.0" } else { "127.0.0.1" }.into();
changed = true;
}
// Friends connect from other addresses, and the server only applies the CoA client protocol to loopback unless remote clients
// are allowed; opening switches that on (closing leaves it, it only matters while the servers listen beyond this computer).
let coa_conf = root.join("Core/configs/modules/coa.conf");
let mut coa_edit = None;
if open {
if let Ok(bytes) = fs::read(&coa_conf) {
let mut conf = ConfFile::parse_bytes(&bytes)?;
if conf.get("CoA.AllowRemoteClients").map(str::trim) != Some("1") {
conf.set("CoA.AllowRemoteClients", "1", &["Set by CoA Server Manager together with the friends mode"]);
originals.push((coa_conf.clone(), bytes));
coa_edit = Some(conf.to_text());
}
}
}
if !changed && coa_edit.is_none() {
return Ok(false);
}
take_snapshot(meta, Scope::Server, if open { "before opening the servers to friends" } else { "before closing the servers to friends" }, &originals)?;
if changed {
fsx::atomic_write_json(&root.join(MARKER), &cfg)?;
}
let remote_changed = coa_edit.is_some();
if let Some(text) = coa_edit {
fsx::atomic_write(&coa_conf, text.as_bytes())?;
}
Ok(changed || remote_changed)
}
/// What each service can be reached from, computed from how the containers publish their ports: the database is never published,
/// the console only on the loopback address, the game servers on `bindAddress`.
pub fn exposure(root: &Path, ports: &Ports) -> Vec<Exposure> {
let open = bind_is_open(root);
let observed = crate::process::observe(root, ports);
let up = |s: &ServiceStatus| matches!(s.state, ServiceState::Running | ServiceState::Starting);
vec![
Exposure { port: ports.mysql, what: "database", listening: up(&observed.mysql), reachable_from_network: false },
Exposure { port: ports.ra, what: "server console", listening: up(&observed.world), reachable_from_network: false },
Exposure { port: ports.auth, what: "login server", listening: up(&observed.auth), reachable_from_network: open && up(&observed.auth) },
Exposure { port: ports.world, what: "game world", listening: up(&observed.world), reachable_from_network: open && up(&observed.world) },
]
}
#[cfg(test)]
mod tests {
use super::*;
fn server(bind: Option<&str>) -> (tempfile::TempDir, PathBuf, PathBuf) {
let d = tempfile::tempdir().unwrap();
let root = d.path().join("srv");
let meta = d.path().join("srv.manager");
fs::create_dir_all(root.join("Settings")).unwrap();
fs::create_dir_all(root.join("Core/configs/modules")).unwrap();
fs::create_dir_all(&meta).unwrap();
let extra = bind.map(|b| format!(r#","bindAddress":"{b}""#)).unwrap_or_default();
fs::write(root.join(MARKER), format!(r#"{{"project":"t1"{extra}}}"#)).unwrap();
fs::write(root.join("Core/configs/modules/coa.conf"), "CoA.Enable = 1\n").unwrap();
(d, root, meta)
}
#[test]
fn opening_and_closing_changes_the_published_address_and_allows_remote_clients() {
let (_d, root, meta) = server(None);
assert!(!bind_is_open(&root));
assert!(set_open(&root, &meta, true).unwrap(), "a restart is needed");
assert!(bind_is_open(&root));
assert_eq!(Config::load(&root).unwrap().bind_address, "0.0.0.0");
assert!(fs::read_to_string(root.join("Core/configs/modules/coa.conf")).unwrap().contains("CoA.AllowRemoteClients = 1"));
assert!(!set_open(&root, &meta, true).unwrap(), "nothing left to change");
assert!(set_open(&root, &meta, false).unwrap());
assert!(!bind_is_open(&root));
assert_eq!(Config::load(&root).unwrap().bind_address, "127.0.0.1");
}
#[test]
fn enabling_remote_client_protocol_requires_restart_even_when_ports_are_open() {
let (_d, root, meta) = server(Some("0.0.0.0"));
fs::write(root.join("Core/configs/modules/coa.conf"), "CoA.Enable = 1\nCoA.AllowRemoteClients = 0\n").unwrap();
assert!(set_open(&root, &meta, true).unwrap());
assert_eq!(Config::load(&root).unwrap().bind_address, "0.0.0.0");
assert!(!set_open(&root, &meta, true).unwrap());
}
#[test]
fn an_address_chosen_by_hand_is_kept_when_opening_and_other_settings_survive() {
let (_d, root, meta) = server(Some("192.168.1.20"));
assert!(bind_is_open(&root), "any address but the loopback is the network");
set_open(&root, &meta, true).unwrap();
let cfg = Config::load(&root).unwrap();
assert_eq!(cfg.bind_address, "192.168.1.20");
assert_eq!(cfg.project, "t1");
}
#[test]
fn the_old_settings_are_snapshotted_before_a_change() {
let (_d, root, meta) = server(None);
set_open(&root, &meta, true).unwrap();
assert_eq!(crate::config::list_snapshots(&meta).len(), 1);
}
#[test]
fn the_database_and_the_console_are_never_reported_reachable() {
let (_d, root, _meta) = server(Some("0.0.0.0"));
let rows = exposure(&root, &Ports::default());
let by = |what: &str| rows.iter().find(|e| e.what == what).unwrap();
assert!(!by("database").reachable_from_network && !by("server console").reachable_from_network);
// Nothing runs in this test (docker is not asked to start anything), so nothing is reachable yet.
assert!(rows.iter().all(|e| !e.reachable_from_network));
}
}
+110 -6
View File
@@ -90,7 +90,12 @@ fn launcher(root: &Path) -> Result<(PathBuf, PathBuf)> {
/// Run a launcher verb to completion (blocking; call from a worker thread).
pub fn run(root: &Path, verb: Verb) -> Result<DriverOutcome> {
run_inner(root, verb, false)
tracing::info!(?verb, root = %root.display(), "driver: launcher requested");
let result = run_inner(root, verb, false);
if let Err(error) = &result {
tracing::warn!(?verb, root = %root.display(), %error, "driver: launcher request failed before completion");
}
result
}
pub(crate) fn validate_update(root: &Path) -> Result<DriverOutcome> {
@@ -100,7 +105,7 @@ pub(crate) fn validate_update(root: &Path) -> Result<DriverOutcome> {
fn run_inner(root: &Path, verb: Verb, validating_update: bool) -> Result<DriverOutcome> {
let root = fsx::canonicalize_lenient(root)?;
let _update_lock = if !validating_update && matches!(verb, Verb::StartAll | Verb::StartWorld) {
Some(crate::update::operation_lock(&crate::registry::metadata_dir_for(&root)?)?)
Some(crate::update::operation_lock_for(&crate::registry::metadata_dir_for(&root)?, "starting server", true)?)
} else { None };
if !validating_update && matches!(verb, Verb::StartAll | Verb::StartWorld) {
crate::update::ensure_recovered(&crate::registry::metadata_dir_for(&root)?)?;
@@ -139,6 +144,7 @@ fn run_inner(root: &Path, verb: Verb, validating_update: bool) -> Result<DriverO
}
}
discard_damaged_launcher_state(&root);
if matches!(verb, Verb::StopAll | Verb::StartMysql | Verb::StartAll) { recover_mysql_launcher_state(&root, verb != Verb::StopAll)?; }
let mut cmd = Command::new(&python);
cmd.arg("-B").arg("-c").arg(LAUNCH_SCRIPT).arg(&script).arg(verb.arg())
.current_dir(&root).stdin(Stdio::null()).stdout(Stdio::piped()).stderr(Stdio::piped());
@@ -175,7 +181,7 @@ fn run_inner(root: &Path, verb: Verb, validating_update: bool) -> Result<DriverO
};
let mut output = out_t.join().unwrap_or_default();
output.push_str(&err_t.join().unwrap_or_default());
let output = output.trim().to_string();
let mut output = output.trim().to_string();
let (ok, exit_code) = match status {
Some(st) => (st.success(), st.code()),
None => (false, None),
@@ -190,6 +196,14 @@ fn run_inner(root: &Path, verb: Verb, validating_update: bool) -> Result<DriverO
other => Some(other),
}
};
if code == Some(ErrorCode::PortInUse) {
let observed = crate::process::observe(&root, &crate::layout::read_ports(&root));
for service in [&observed.mysql, &observed.auth, &observed.world].into_iter().chain(observed.secondary_world.iter()) {
if let Some(conflict) = &service.conflict {
output.push_str(&format!("\nPort {} is used by process {} ({}).", conflict.port, conflict.pid, conflict.exe.as_deref().unwrap_or("executable unavailable")));
}
}
}
tracing::info!(?verb, ok, ?code, "driver: launcher verb finished");
if !ok && !output.is_empty() {
let redacted = crate::diag::redact(&output);
@@ -240,7 +254,37 @@ pub(crate) fn discard_damaged_launcher_state(root: &Path) -> Vec<PathBuf> {
removed
}
const LAUNCH_SCRIPT: &str = "import runpy,sys;from pathlib import Path;script=sys.argv[1];sys.path.insert(0,str(Path(script).resolve().parent));sys.argv=sys.argv[1:];runpy.run_path(script,run_name='__main__')";
/// A lost launcher record must not hide this installation's running MySQL from
/// graceful shutdown. Adopt only the exact executable owning its configured port.
fn recover_mysql_launcher_state(root: &Path, prefer_listener: bool) -> Result<()> {
let observed = crate::process::observe(root, &crate::layout::read_ports(root));
if observed.mysql.state != crate::process::ServiceState::Running || !observed.mysql.port_ready { return Ok(()); }
let Some(pid) = observed.mysql.pid else { return Ok(()); };
let Some(identity) = crate::process::identity(pid) else { return Ok(()); };
let (_, _, executable) = crate::layout::executables(root);
if !crate::process::find_by_exe(&executable).iter().any(|candidate| candidate.same_as(&identity))
|| !crate::process::listeners().contains(&(observed.mysql.port, pid)) { return Ok(()); }
if let Some(saved) = crate::process::read_state_record(root, "mysql") {
if saved.same_as(&identity) { return Ok(()); }
if crate::process::is_alive(&saved) {
// A valid record may describe the MySQL supervisor rather than its
// listening child. The launcher can gracefully stop either record.
if crate::process::find_by_exe(&executable).iter().any(|candidate| candidate.same_as(&saved)) {
if !prefer_listener { return Ok(()); }
} else {
return Err(Error::Invalid("The MySQL launcher record identifies a different live process. No database action was requested.".into()));
}
}
}
let state = root.join(".state/mysql.json");
let previous = root.join(".state/mysql-before-shutdown-recovery.json");
if state.is_file() && !previous.exists() { fsx::atomic_write(&previous, &std::fs::read(&state)?)?; }
fsx::atomic_write_json(&state, &identity)?;
tracing::warn!(pid, root = %root.display(), "Recovered MySQL launcher identity from the installed executable owning its port");
Ok(())
}
const LAUNCH_SCRIPT: &str = "import runpy,sys\nfrom pathlib import Path\nscript=sys.argv[1]\nsys.path.insert(0,str(Path(script).resolve().parent))\nsys.argv=sys.argv[1:]\nif sys.argv[1]=='stop-all':\n launcher=runpy.run_path(script,run_name='coa_manager_shutdown')\n config=launcher['settings']()\n with launcher['control_lock']():\n launcher['stop_all'](config)\nelse:\n runpy.run_path(script,run_name='__main__')";
pub(crate) fn patch_launcher_imports(text: &str) -> Result<String> {
let line = "sys.path.insert(0, str(Path(__file__).resolve().parent))";
@@ -279,14 +323,62 @@ pub(crate) fn startup_failure(root: &Path, started: &DriverOutcome) -> String {
let title = code.map(|c| c.human().title).unwrap_or("The server did not become ready");
let output = crate::diag::redact(&started.output);
let lines: Vec<_> = output.lines().rev().take(12).collect();
if lines.is_empty() { title.into() }
else { format!("{title}\n{}", lines.into_iter().rev().collect::<Vec<_>>().join("\n")) }
let mut reason = if lines.is_empty() { title.into() }
else { format!("{title}\n{}", lines.into_iter().rev().collect::<Vec<_>>().join("\n")) };
for relative in ["Core/Logs/world-console.log", "Core/Logs/auth-console.log", "mysql/logs/mysql-error.log"] {
let tail = crate::health::tail(&root.join(relative), 64 * 1024);
let failures: Vec<_> = tail.lines().filter(|line| {
let lower = line.to_ascii_lowercase();
crate::health::diagnose(line).is_some() || lower.contains("fatal") || lower.contains("error") || lower.contains("aborting")
}).rev().take(8).collect();
if !failures.is_empty() {
reason.push_str(&format!("\n{relative}:\n{}", crate::diag::redact(&failures.into_iter().rev().collect::<Vec<_>>().join("\n"))));
}
}
reason
}
#[cfg(test)]
mod tests {
use super::*;
#[cfg(windows)]
#[test]
fn shutdown_recovery_never_adopts_a_foreign_port_listener() {
let root = tempfile::tempdir().unwrap();
let listener = std::net::TcpListener::bind("127.0.0.1:0").unwrap();
fsx::atomic_write_json(&root.path().join("Settings/repack.json"), &serde_json::json!({"mysqlPort":listener.local_addr().unwrap().port()})).unwrap();
recover_mysql_launcher_state(root.path(), false).unwrap();
recover_mysql_launcher_state(root.path(), true).unwrap();
assert!(!root.path().join(".state/mysql.json").exists());
assert!(crate::process::identity(std::process::id()).is_some());
}
#[test]
fn a_blocked_start_is_logged_before_any_launcher_is_spawned() {
use std::io::Write;
use std::sync::{Arc, Mutex};
struct Capture(Arc<Mutex<Vec<u8>>>);
impl Write for Capture {
fn write(&mut self, bytes: &[u8]) -> std::io::Result<usize> { self.0.lock().unwrap().extend_from_slice(bytes); Ok(bytes.len()) }
fn flush(&mut self) -> std::io::Result<()> { Ok(()) }
}
let temp = tempfile::tempdir().unwrap();
let root = temp.path().join("server");
std::fs::create_dir(&root).unwrap();
let metadata = crate::registry::metadata_dir_for(&root).unwrap();
let _maintenance = crate::update::operation_lock(&metadata).unwrap();
let buffer = Arc::new(Mutex::new(Vec::new()));
let writer = Arc::clone(&buffer);
let subscriber = tracing_subscriber::fmt().without_time().with_ansi(false).with_writer(move || Capture(Arc::clone(&writer))).finish();
let error = tracing::subscriber::with_default(subscriber, || run(&root, Verb::StartAll)).unwrap_err();
assert!(error.to_string().contains("in progress"));
let text = String::from_utf8(buffer.lock().unwrap().clone()).unwrap();
assert!(text.contains("launcher requested") && text.contains("StartAll"));
assert!(text.contains("launcher request failed before completion") && text.contains("in progress"));
assert!(!text.contains("starting launcher verb"));
}
#[test]
fn background_start_disables_cli_in_template_and_rendered_config() {
let folder = tempfile::tempdir().unwrap();
@@ -381,6 +473,18 @@ mod tests {
assert!(run(dir.path(), Verb::StartAll).is_err());
}
#[test]
fn startup_failure_reports_late_world_errors_after_launcher_ready() {
let dir = tempfile::tempdir().unwrap();
std::fs::create_dir_all(dir.path().join("Core/Logs")).unwrap();
std::fs::write(dir.path().join("Core/Logs/world-console.log"), "ERROR [1146] Table 'acore_characters.playerbots_names' doesn't exist\nFATAL Could not prepare statements\n").unwrap();
let started = DriverOutcome { ok: true, code: None, exit_code: Some(0), human: None, output: "Worldserver and automatic bug reporting are ready.".into() };
let reason = startup_failure(dir.path(), &started);
assert!(reason.contains("playerbots_names"));
assert!(reason.contains("Core/Logs/world-console.log"));
assert!(reason.contains("Could not prepare statements"));
}
#[test]
fn startup_failure_keeps_the_launcher_reason_without_secrets() {
let dir = tempfile::tempdir().unwrap();
+6
View File
@@ -55,6 +55,7 @@ pub enum ErrorCode {
OperationInProgress,
StartupFailed,
DockerUnavailable,
GameDataMismatch,
HashMismatch,
PathRejected,
PackageNotPublished,
@@ -138,6 +139,11 @@ impl ErrorCode {
"This server runs in Docker. Install Docker, start it, and make sure your user account may use it, then try again.",
&[Retry, ShowDetails],
),
ErrorCode::GameDataMismatch => (
"The game data folder is not the CoA set",
"The server stopped because the DBC files in the game data folder are not the ones of the Conquest of Azeroth client. Choose the folder that holds the CoA data (Settings > Game data folder), then start again.",
&[ShowDetails],
),
ErrorCode::HashMismatch => (
"A downloaded file is damaged",
"The file did not match its checksum and was not used.",
+9 -2
View File
@@ -80,6 +80,9 @@ fn bind_targets(root: &Path) -> Vec<PathBuf> {
/// Whether the configuration currently lets other computers reach the login and world servers.
pub fn bind_is_open(root: &Path) -> bool {
if crate::docker::is_docker(root) {
return crate::docker::bind_is_open(root);
}
let files = bind_targets(root);
!files.is_empty()
&& files
@@ -90,6 +93,9 @@ pub fn bind_is_open(root: &Path) -> bool {
/// Open (`0.0.0.0`) or close (`127.0.0.1`) the login and world servers to other computers. The database and the server
/// console are never touched. Returns true if anything changed (a restart is then needed). The old files are snapshotted.
pub fn set_open(root: &Path, meta: &Path, open: bool) -> Result<bool> {
if crate::docker::is_docker(root) {
return crate::docker::set_open(root, meta, open);
}
let want = if open { "\"0.0.0.0\"" } else { "\"127.0.0.1\"" };
let mut originals = Vec::new();
let mut edits = Vec::new();
@@ -125,7 +131,8 @@ pub fn set_open(root: &Path, meta: &Path, open: bool) -> Result<bool> {
Ok(true)
}
/// Make the realm list advertise `host` (the launcher resets it to 127.0.0.1 at every start, so this runs after each start).
/// Make the realm list advertise `host` to remote clients (the launcher resets it to 127.0.0.1 at every start, so this
/// runs after each start).
pub fn apply_realm_address(root: &Path, host: &str) -> Result<()> {
if !host_ok(host) {
return Err(Error::Invalid("That address is not valid.".into()));
@@ -133,7 +140,7 @@ pub fn apply_realm_address(root: &Path, host: &str) -> Result<()> {
let db = Db::from_repack(root, Account::Admin)?;
let realm = crate::realms::state(root)?.active.realm_id();
let where_realms = if crate::realms::state(root)?.simultaneous { "id IN (1,2)".into() } else { format!("id={realm}") };
db.query(&format!("UPDATE acore_auth.realmlist SET address='{host}', localAddress='{host}' WHERE {where_realms};"))?;
db.query(&format!("UPDATE acore_auth.realmlist SET address='{host}' WHERE {where_realms};"))?;
Ok(())
}
+42
View File
@@ -28,6 +28,19 @@ pub fn normalize_lexical(path: &Path) -> PathBuf {
fn key(path: &Path) -> Vec<String> {
path.components()
.map(|c| {
#[cfg(windows)]
if let Component::Prefix(prefix) = c {
use std::path::Prefix;
match prefix.kind() {
Prefix::Disk(drive) | Prefix::VerbatimDisk(drive) => {
return format!("{}:", char::from(drive).to_ascii_lowercase());
}
Prefix::UNC(server, share) | Prefix::VerbatimUNC(server, share) => {
return format!("\\\\{}\\{}", server.to_string_lossy(), share.to_string_lossy()).to_lowercase();
}
_ => {}
}
}
let c = c.as_os_str().to_string_lossy();
if cfg!(windows) { c.to_lowercase() } else { c.into_owned() }
})
@@ -262,6 +275,35 @@ mod tests {
}
}
#[cfg(windows)]
#[test]
fn containment_matches_verbatim_drive_and_unc_prefixes() {
assert!(starts_with_ci(Path::new(r"\\?\C:\server\Extras\file.sql"), Path::new(r"C:\server")));
assert!(starts_with_ci(Path::new(r"C:\server\Extras\file.sql"), Path::new(r"\\?\c:\server")));
assert!(starts_with_ci(Path::new(r"\\?\UNC\host\share\server\file.sql"), Path::new(r"\\host\share\server")));
assert!(!starts_with_ci(Path::new(r"\\?\D:\server\file.sql"), Path::new(r"C:\server")));
assert!(!starts_with_ci(Path::new(r"\\?\C:\server-other\file.sql"), Path::new(r"C:\server")));
assert!(!starts_with_ci(Path::new(r"\\?\UNC\host\other\server\file.sql"), Path::new(r"\\host\share\server")));
}
#[cfg(windows)]
#[test]
fn ensure_within_accepts_long_rehearsal_sql_path() {
let dir = tempfile::tempdir().unwrap();
let root = dir.path().join("server");
fs::create_dir_all(&root).unwrap();
let mut candidate = root.clone();
for _ in 0..5 {
candidate.push("long-private-rehearsal-directory-name-20261013");
}
fs::create_dir_all(&candidate).unwrap();
candidate.push("2026_06_13_00_ai_playerbot_missing_group_buff_reagent.sql");
fs::write(&candidate, b"SELECT 1;").unwrap();
assert!(candidate.as_os_str().len() > 260);
assert!(ensure_within(&root, &candidate).is_ok());
assert!(ensure_within(&root, &std::fs::canonicalize(candidate).unwrap()).is_ok());
}
#[test]
fn ensure_within_blocks_junction_escape() {
if !cfg!(windows) {
+9
View File
@@ -25,6 +25,8 @@ pub fn diagnose(text: &str) -> Option<ErrorCode> {
Some(ErrorCode::DatabaseNotRunning)
} else if l.contains("address already in use") || l.contains("could not bind") || l.contains("only one usage of each socket") {
Some(ErrorCode::PortInUse)
} else if l.contains("does not hold the coa client dbc set") || l.contains("_outdated_ dbc data") {
Some(ErrorCode::GameDataMismatch)
} else if l.contains("config::loadfile") && l.contains("failed open file") && l.contains("worldserver.conf") {
Some(ErrorCode::ServerFilesIncomplete)
} else {
@@ -57,6 +59,13 @@ pub fn diagnose_installation(root: &Path) -> Option<ErrorCode> {
mod tests {
use super::*;
#[test]
fn a_game_data_folder_without_the_coa_dbc_set_is_named_as_such() {
let log = "Using DataDir /srv/data/\nSpellAffect.dbc not found in /srv/data/dbc/\nDataDir does not hold the CoA client DBC set. Install it with apps/coa-dbc/client_dbc.py.";
assert_eq!(diagnose(log), Some(ErrorCode::GameDataMismatch));
assert_eq!(diagnose("You have _outdated_ DBC data. Please extract correct versions from current using client."), Some(ErrorCode::GameDataMismatch));
}
#[test]
fn recognises_missing_migrations() {
let log = "ERROR [sql.sql] [1146] Table 'acore_world.coa_boss' doesn't exist\nFATAL [sql.sql] Your database structure is not up to date.";
+4
View File
@@ -201,6 +201,10 @@ fn bootstrap_database_inner(root: &Path) -> Result<()> {
sql.push_str("FLUSH PRIVILEGES;\n");
db.query(&sql)?;
fsx::atomic_write_json(&target, &serde_json::json!({ "rootPassword": root_pw, "appPassword": app_pw }))?;
fsx::atomic_write(&root.join("mysql/admin-client.ini"), format!(
"[client]\nuser=root\npassword={root_pw}\nhost=127.0.0.1\nport={}\nprotocol=tcp\ndefault-character-set=utf8mb4\n",
crate::layout::read_ports(root).mysql,
).as_bytes())?;
// The server console gets its own account with a random password instead of a shared default.
let console_pw = Db::from_repack(root, Account::Admin)?.provision_service_account()?; // new root password
db::write_console_credentials(root, &console_pw)?;
+1
View File
@@ -44,6 +44,7 @@ pub mod release;
pub mod report;
pub mod release_schema;
pub mod update;
pub mod update_route;
mod update_isolation;
mod mysql_snapshot;
pub mod wine;
+1
View File
@@ -93,6 +93,7 @@ pub struct Compatibility {
pub core_commit: String,
pub module_commits: std::collections::BTreeMap<String, String>,
pub client_patch_version: String,
/// Sources exercised during release qualification, not an exclusive updater allowlist.
pub source_versions: Vec<String>,
#[serde(default, skip_serializing_if = "std::collections::BTreeMap::is_empty")]
pub source_databases: std::collections::BTreeMap<String, SourceDatabase>,
+148 -52
View File
@@ -11,6 +11,7 @@ use crate::fsx;
use crate::manifest::Migration;
const LEDGER_SCHEMA: &str = "acore_world";
const AUTH_LEDGER_SCHEMA: &str = "acore_auth";
const LEDGER_TABLE: &str = "coa_manager_migrations";
#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize)]
@@ -37,6 +38,10 @@ pub trait Store {
fn load(&self) -> Result<Vec<LedgerRow>>;
fn load_readonly(&self) -> Result<Vec<LedgerRow>> { self.load() }
fn put(&self, row: &LedgerRow) -> Result<()>;
fn put_batch(&self, rows: &[LedgerRow]) -> Result<()> {
for row in rows { self.put(row)?; }
Ok(())
}
fn run_file(&self, schema: &str, file: &Path) -> Result<()>;
}
@@ -48,59 +53,101 @@ fn ident_ok(s: &str) -> bool {
!s.is_empty() && s.len() <= 150 && s.chars().all(|c| c.is_ascii_alphanumeric() || matches!(c, '_' | '.' | '-'))
}
pub(crate) fn shared_auth(kind: &str) -> bool {
matches!(db::schema_of(kind), Ok("acore_auth"))
}
fn create_ledger(db: &Db, schema: &str) -> Result<()> {
db.query(&format!("CREATE TABLE IF NOT EXISTS `{schema}`.`{LEDGER_TABLE}` (`db` VARCHAR(32) NOT NULL, `id` VARCHAR(190) NOT NULL, `sha256` CHAR(64) NOT NULL, `status` VARCHAR(16) NOT NULL, `applied_at` DATETIME NOT NULL, `error` TEXT NULL, `baseline` TINYINT NOT NULL DEFAULT 0, PRIMARY KEY (`db`,`id`)) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;"))?;
Ok(())
}
impl Store for Db {
fn load(&self) -> Result<Vec<LedgerRow>> {
self.query(&format!(
"CREATE TABLE IF NOT EXISTS `{LEDGER_SCHEMA}`.`{LEDGER_TABLE}` (`db` VARCHAR(32) NOT NULL, `id` VARCHAR(190) NOT NULL, `sha256` CHAR(64) NOT NULL, `status` VARCHAR(16) NOT NULL, `applied_at` DATETIME NOT NULL, `error` TEXT NULL, `baseline` TINYINT NOT NULL DEFAULT 0, PRIMARY KEY (`db`,`id`)) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;"
))?;
self.load_readonly()
create_ledger(self, LEDGER_SCHEMA)?;
create_ledger(self, AUTH_LEDGER_SCHEMA)?;
let rows = self.load_readonly()?;
// Preserve already-recorded core history even if a module later replaces its updates table.
// A zero checksum remains unknown; importing it does not invent a checksum for old SQL.
let selected: std::collections::BTreeMap<_, _> = rows.iter().map(|row| ((row.db.clone(), row.id.to_ascii_lowercase()), row)).collect();
let core: Vec<_> = selected.values().filter(|r| r.status == Status::Applied && r.baseline && r.sha256 == "0".repeat(64)
&& matches!(r.db.as_str(), "world" | "characters") && ident_ok(&r.id)).map(|r| (*r).clone()).collect();
self.put_batch(&core)?;
Ok(rows)
}
fn load_readonly(&self) -> Result<Vec<LedgerRow>> {
let out = if self.tables(LEDGER_SCHEMA)?.iter().any(|name| name == LEDGER_TABLE) {
self.query(&format!("SELECT `db`,`id`,`sha256`,`status`,`baseline`,COALESCE(HEX(`error`),'') FROM `{LEDGER_SCHEMA}`.`{LEDGER_TABLE}` ORDER BY `db`,`id`;"))?
} else { String::new() };
let mut rows = Vec::new();
// Files the core's own updater (or whoever prepared this database) already recorded count as applied.
for (kind, schema) in db::SCHEMAS {
if self.tables(schema)?.iter().any(|name| name == "updates") {
let names = self.query(&format!("SELECT name FROM `{schema}`.`updates`;"))?;
for name in names.lines().filter(|n| n.ends_with(".sql")) {
rows.push(LedgerRow { db: kind.into(), id: name.trim_end_matches(".sql").into(), sha256: "0".repeat(64), status: Status::Applied, error: None, baseline: true });
}
// Resolve physical schemas once, then read all history in two client calls.
// Using a neutral connection prevents a Wildcard query from rerouting the other realm.
let connection = self.clone().for_realm(crate::realms::Mode::Coa);
let world = self.realm_schema("acore_world");
let characters = self.realm_schema("acore_characters");
let other = if self.realm() == crate::realms::Mode::Coa { "acore_world_wildcard" } else { "acore_world" };
let schemas: std::collections::BTreeSet<_> = [world, characters, other, AUTH_LEDGER_SCHEMA].into_iter().collect();
let catalog = connection.query(&format!("SELECT table_schema,table_name FROM information_schema.tables WHERE table_schema IN ('{}') AND table_name IN ('updates','{LEDGER_TABLE}') AND table_type='BASE TABLE';", schemas.into_iter().collect::<Vec<_>>().join("','")))?;
let available: std::collections::BTreeSet<_> = catalog.lines().filter_map(|line| line.split_once('\t')).collect();
let text = |expression: &str| format!("CONVERT({expression} USING utf8mb4) COLLATE utf8mb4_bin");
let mut queries = Vec::new();
for (kind, schema) in [("characters", characters), ("auth", AUTH_LEDGER_SCHEMA), ("world", world)] {
if available.contains(&(schema, "updates")) {
queries.push(format!("SELECT 'core','{kind}',{},REPEAT('0',64),'applied',1,'' FROM `{schema}`.`updates` WHERE BINARY RIGHT(name,4)='.sql'",text("LEFT(name,CHAR_LENGTH(name)-4)")));
}
}
for line in out.lines().filter(|l| !l.is_empty()) {
let f: Vec<&str> = line.split('\t').collect();
if f.len() < 5 {
return Err(Error::Invalid("The database migration history is malformed; no SQL was replayed.".into()));
for (origin, schema, filter) in [("local",world,""), ("other",other," WHERE `db` IN ('auth','coa-auth','wildcard-auth')"), ("shared",AUTH_LEDGER_SCHEMA,"")] {
if available.contains(&(schema, LEDGER_TABLE)) {
queries.push(format!("SELECT '{origin}',{},{},{},{},`baseline`,{} FROM `{schema}`.`{LEDGER_TABLE}`{filter}",text("`db`"),text("`id`"),text("`sha256`"),text("`status`"),text("COALESCE(HEX(`error`),'')")));
}
if !matches!(f[3], "applied" | "failed" | "running") || !ident_ok(f[1]) || f[2].len() != 64 || !f[2].chars().all(|c| c.is_ascii_hexdigit()) {
}
if queries.is_empty() { return Ok(vec![]); }
let out = connection.query(&queries.join(" UNION ALL "))?;
let (mut rows, mut local, mut auth, mut shared) = (Vec::new(), Vec::new(), Vec::new(), Vec::new());
for line in out.lines().filter(|l| !l.is_empty()) {
let f: Vec<_> = line.split('\t').collect();
if f.len() < 6 { return Err(Error::Invalid("The database migration history is malformed; no SQL was replayed.".into())); }
// Unrelated legacy filenames cannot match a valid manifest migration ID.
if f[0] == "core" && !ident_ok(f[2]) { continue; }
if !matches!(f[4], "applied" | "failed" | "running") || !ident_ok(f[2]) || f[3].len() != 64 || !f[3].chars().all(|c| c.is_ascii_hexdigit()) {
return Err(Error::Invalid("The database migration history is invalid; no SQL was replayed.".into()));
}
let err = f.get(5).filter(|h| !h.is_empty()).and_then(|h| hex::decode(h).ok()).map(|b| String::from_utf8_lossy(&b).into_owned());
rows.push(LedgerRow {
db: f[0].into(),
id: f[1].into(),
sha256: f[2].into(),
status: match f[3] { "applied" => Status::Applied, "running" => Status::Running, _ => Status::Failed },
baseline: f[4] == "1",
error: err,
});
let error = f.get(6).filter(|h| !h.is_empty()).and_then(|h| hex::decode(h).ok()).map(|b| String::from_utf8_lossy(&b).into_owned());
let row = LedgerRow { db:f[1].into(), id:f[2].into(), sha256:f[3].into(), status:match f[4] { "applied"=>Status::Applied,"running"=>Status::Running,_=>Status::Failed }, baseline:f[5]=="1", error };
match f[0] {
"core" => rows.push(row),
"local" if !shared_auth(&row.db) => local.push(row),
"local" | "other" => auth.push(row),
"shared" => shared.push(row),
_ => return Err(Error::Invalid("Unknown migration history source.".into())),
}
}
rows.extend(local);
// An interrupted legacy auth execution must not be hidden by another realm's applied row.
auth.sort_by_key(|r| (match r.status { Status::Applied=>0,Status::Running=>1,Status::Failed=>2,Status::Pending=>3 },r.sha256!="0".repeat(64)));
for mut row in auth { row.db="auth".into(); rows.push(row); }
// New auth records live with the shared account database and are authoritative.
for mut row in shared {
if !shared_auth(&row.db) { return Err(Error::Invalid("The shared authentication migration history is invalid.".into())); }
rows.retain(|r| !(shared_auth(&r.db) && r.id.eq_ignore_ascii_case(&row.id)));
row.db="auth".into(); rows.push(row);
}
Ok(rows)
}
fn put(&self, r: &LedgerRow) -> Result<()> {
if !ident_ok(&r.id) || !ident_ok(&r.db) || r.sha256.len() != 64 || !r.sha256.chars().all(|c| c.is_ascii_hexdigit()) {
return Err(Error::Invalid("migration record has an invalid identifier".into()));
fn put(&self, row: &LedgerRow) -> Result<()> { self.put_batch(std::slice::from_ref(row)) }
fn put_batch(&self, rows: &[LedgerRow]) -> Result<()> {
let mut groups = std::collections::BTreeMap::<&str, Vec<String>>::new();
for r in rows {
if !ident_ok(&r.id) || !ident_ok(&r.db) || r.sha256.len() != 64 || !r.sha256.chars().all(|c| c.is_ascii_hexdigit()) {
return Err(Error::Invalid("migration record has an invalid identifier".into()));
}
let status = match r.status { Status::Applied => "applied", Status::Running => "running", _ => "failed" };
let error = r.error.as_deref().map(|e| format!("UNHEX('{}')", hex_of(e))).unwrap_or_else(|| "NULL".into());
let (schema, kind) = if shared_auth(&r.db) { (AUTH_LEDGER_SCHEMA, "auth") } else { (LEDGER_SCHEMA, r.db.as_str()) };
groups.entry(schema).or_default().push(format!("('{}','{}','{}','{status}',NOW(),{error},{})",kind,r.id,r.sha256,r.baseline as u8));
}
for (schema, values) in groups {
self.query(&format!("REPLACE INTO `{schema}`.`{LEDGER_TABLE}` (`db`,`id`,`sha256`,`status`,`applied_at`,`error`,`baseline`) VALUES {};",values.join(",")))?;
}
let status = match r.status { Status::Applied => "applied", Status::Running => "running", _ => "failed" };
let err = r.error.as_deref().map(|e| format!("UNHEX('{}')", hex_of(e))).unwrap_or_else(|| "NULL".into());
self.query(&format!(
"REPLACE INTO `{LEDGER_SCHEMA}`.`{LEDGER_TABLE}` (`db`,`id`,`sha256`,`status`,`applied_at`,`error`,`baseline`) VALUES ('{}','{}','{}','{status}',NOW(),{err},{});",
r.db, r.id, r.sha256, r.baseline as u8
))?;
Ok(())
}
@@ -137,21 +184,34 @@ pub fn status(store: &dyn Store, list: &[Migration]) -> Result<Vec<Item>> {
.map(|m| {
// Manager records are appended after the core's updates table and are authoritative.
let row = recorded(&rows, m);
let changed = row.is_some_and(|r| hash_changed(r, m));
let history = effective_history(&rows, m);
let changed = history.iter().any(|r| hash_changed(r,m));
let interrupted = history.iter().find(|r| matches!(r.status, Status::Running | Status::Failed));
Item {
id: m.id.clone(),
db: m.db.clone(),
status: if changed || row.is_some_and(|r| r.status == Status::Running) { Status::Failed } else { row.map(|r| r.status).unwrap_or(Status::Pending) },
status: if changed || interrupted.is_some() { Status::Failed } else { row.map(|r| r.status).unwrap_or(Status::Pending) },
destructive: m.destructive,
baseline: row.map(|r| r.baseline).unwrap_or(false),
error: if changed { Some("An already applied SQL file changed. Publish a new migration instead of replaying it.".into()) } else { row.and_then(|r| r.error.clone()) },
error: if changed { Some("An already applied SQL file changed. Publish a new migration instead of replaying it.".into()) } else { interrupted.and_then(|r| r.error.clone()).or_else(|| row.and_then(|r| r.error.clone())) },
}
})
.collect())
}
fn recorded<'a>(rows: &'a [LedgerRow], m: &Migration) -> Option<&'a LedgerRow> {
rows.iter().rev().find(|r| r.db == m.db && r.id == m.id)
rows.iter().rev().find(|r| (r.db == m.db || shared_auth(&r.db) && shared_auth(&m.db)) && r.id.eq_ignore_ascii_case(&m.id))
}
fn effective_history<'a>(rows: &'a [LedgerRow], m: &Migration) -> Vec<&'a LedgerRow> {
if shared_auth(&m.db) { rows.iter().filter(|r| shared_auth(&r.db) && r.id.eq_ignore_ascii_case(&m.id)).collect() }
else { recorded(rows, m).into_iter().collect() }
}
fn preserve_auth_history(store: &dyn Store, rows: &[LedgerRow], list: &[Migration]) -> Result<()> {
let applied: Vec<_> = list.iter().filter(|m| shared_auth(&m.db)).filter_map(|m| recorded(rows,m))
.filter(|r| r.status == Status::Applied).cloned().collect();
store.put_batch(&applied)
}
/// Migrations that were published with different bytes but have the same effect, so a server that applied either
@@ -166,7 +226,7 @@ const SAME_EFFECT: &[(&str, &[&str])] = &[(
)];
fn same_effect(id: &str, a: &str, b: &str) -> bool {
SAME_EFFECT.iter().any(|(known, hashes)| *known == id
SAME_EFFECT.iter().any(|(known, hashes)| known.eq_ignore_ascii_case(id)
&& hashes.iter().any(|h| h.eq_ignore_ascii_case(a)) && hashes.iter().any(|h| h.eq_ignore_ascii_case(b)))
}
@@ -183,8 +243,7 @@ pub fn preflight(store: &dyn Store, list: &[Migration]) -> Result<()> {
pub fn pending_count(store: &dyn Store, list: &[Migration]) -> Result<usize> {
let rows = store.load_readonly()?;
check_history(&rows, list)?;
Ok(list.iter().filter(|m| recorded(&rows, m).is_none_or(|r| r.status != Status::Applied)).count())
Ok(pending_migrations(&rows, list)?.len())
}
fn verified_file(m: &Migration, dir: &Path) -> Result<(&'static str, std::path::PathBuf, bool)> {
@@ -199,22 +258,39 @@ fn verified_file(m: &Migration, dir: &Path) -> Result<(&'static str, std::path::
/// Check every migration artifact, including applied history, before replacing server files.
pub fn verify_files(list: &[Migration], dir: &Path) -> Result<()> {
validate_list(list)?;
for m in list { verified_file(m, dir)?; }
Ok(())
}
fn validate_list(list: &[Migration]) -> Result<()> {
let mut seen = std::collections::BTreeSet::new();
for m in list {
if !seen.insert((&m.db, &m.id)) { return Err(Error::InvalidManifest(format!("duplicate migration {} ({})", m.id, m.db))); }
verified_file(m, dir)?;
if !ident_ok(&m.id) { return Err(Error::InvalidManifest(format!("migration id {:?} is not allowed", m.id))); }
let schema = db::schema_of(&m.db)?;
if !seen.insert((schema, m.id.to_ascii_lowercase())) {
return Err(Error::InvalidManifest(format!("duplicate migration {} ({})", m.id, m.db)));
}
}
Ok(())
}
fn pending_migrations<'a>(rows: &[LedgerRow], list: &'a [Migration]) -> Result<Vec<&'a Migration>> {
check_history(rows, list)?;
Ok(list.iter().filter(|m| recorded(rows, m).is_none_or(|r| r.status != Status::Applied)).collect())
}
fn check_history(rows: &[LedgerRow], list: &[Migration]) -> Result<()> {
validate_list(list)?;
for m in list {
if recorded(rows, m).is_some_and(|r| matches!(r.status, Status::Running | Status::Failed)) {
for row in effective_history(rows,m) {
if matches!(row.status, Status::Running | Status::Failed) {
return Err(Error::Invalid(format!("Database update {} ({}) failed or was interrupted and may be partially applied. Restore its recovery point before retrying.", m.id, m.db)));
}
if recorded(rows, m).is_some_and(|r| hash_changed(r, m)) {
if hash_changed(row,m) {
return Err(Error::Invalid(format!("Applied database update {} ({}) has a different checksum. A new migration is required; it was not replayed.", m.id, m.db)));
}
}
}
Ok(())
}
@@ -223,9 +299,11 @@ fn check_history(rows: &[LedgerRow], list: &[Migration]) -> Result<()> {
/// was brought up to date by hand).
pub fn baseline(store: &dyn Store, list: &[Migration]) -> Result<usize> {
let rows = store.load()?;
check_history(&rows,list)?;
preserve_auth_history(store,&rows,list)?;
let mut n = 0;
for m in list {
if rows.iter().any(|r| r.db == m.db && r.id == m.id && r.status == Status::Applied) {
if recorded(&rows,m).is_some_and(|r| r.status == Status::Applied) {
continue;
}
store.put(&LedgerRow { db: m.db.clone(), id: m.id.clone(), sha256: m.sha256.clone(), status: Status::Applied, error: None, baseline: true })?;
@@ -246,8 +324,7 @@ pub struct ApplyReport {
pub fn apply_pending(store: &dyn Store, list: &[Migration], dir: &Path, snapshot: &dyn Fn() -> Result<String>) -> Result<ApplyReport> {
let rows = store.load()?;
// The list order is the release's order of application (released updates, then pending, then modules).
check_history(&rows, list)?;
let todo: Vec<&Migration> = list.iter().filter(|m| recorded(&rows, m).is_none_or(|r| r.status != Status::Applied)).collect();
let todo = pending_migrations(&rows, list)?;
tracing::info!(total = list.len(), pending = todo.len(), skipped = list.len() - todo.len(), "database migration plan");
// Resolve and verify every file before running anything.
@@ -261,11 +338,13 @@ pub fn apply_pending(store: &dyn Store, list: &[Migration], dir: &Path, snapshot
let mut report = ApplyReport { applied: Vec::new(), failed: None, snapshot: None };
if files.is_empty() {
preserve_auth_history(store,&rows,list)?;
return Ok(report);
}
if any_destructive {
report.snapshot = Some(snapshot().map_err(|e| Error::Invalid(format!("No database update was applied because the safety backup failed: {e}")))?);
}
preserve_auth_history(store,&rows,list)?;
for (m, schema, path) in files {
tracing::info!(migration = %m.id, database = %m.db, "database migration starting");
store.put(&LedgerRow { db: m.db.clone(), id: m.id.clone(), sha256: m.sha256.clone(), status: Status::Running, error: Some("Interrupted SQL must be recovered before replay.".into()), baseline: false })?;
@@ -368,6 +447,23 @@ mod tests {
assert!(store.ran.borrow().is_empty());
}
#[test]
fn every_entry_point_rejects_duplicate_sql_before_execution() {
let (dir, original) = setup(&[("world", "once", "SELECT 1;\n")]);
for alias in ["world", "coa-world"] {
let mut list = original.clone();
let mut duplicate = original[0].clone();
duplicate.db = alias.into();
list.push(duplicate);
let store = Mem::default();
assert!(preflight(&store, &list).is_err());
assert!(pending_count(&store, &list).is_err());
assert!(apply_pending(&store, &list, dir.path(), &no_snapshot).is_err());
assert!(store.ran.borrow().is_empty());
assert!(store.rows.borrow().is_empty());
}
}
#[test]
fn staged_files_reject_missing_corrupt_duplicate_and_invalid_migrations() {
let (dir, list) = setup(&[("characters", "repair", "SELECT 1;\n")]);
+73
View File
@@ -119,7 +119,26 @@ fn conf_name(root: &Path, e: &Entry) -> String {
}
}
fn installed_module_commits(root: &Path) -> BTreeMap<String, String> {
let Ok(directory) = crate::registry::metadata_dir_for(root) else { return Default::default() };
let Ok((_, meta)) = crate::registry::MetaDir::open(&directory) else { return Default::default() };
let Some(version) = meta.core.version else { return Default::default() };
let candidates = [directory.join("manifests").join(format!("update-{version}.json")), directory.join("manifests/base.json")];
for path in candidates {
let Ok(manifest) = crate::fsx::read_json::<serde_json::Value>(&path) else { continue };
if manifest.get("version").and_then(|value| value.as_str()) != Some(&version) { continue; }
if let Some(commits) = manifest.get("compatibility").and_then(|value| value.get("moduleCommits")).and_then(|value| value.as_object()) {
return commits.iter().filter_map(|(name, commit)| {
let commit = commit.as_str()?;
(commit.len() == 40 && commit.bytes().all(|byte| byte.is_ascii_hexdigit())).then(|| (name.clone(), commit.chars().take(8).collect()))
}).collect();
}
}
Default::default()
}
fn list_in(cat: &[Entry], root: &Path) -> Vec<ModuleView> {
let installed_commits = installed_module_commits(root);
let wildcard = crate::realms::state(root).map(|s| s.active == crate::realms::Mode::Wildcard).unwrap_or(true);
cat.iter()
.cloned()
@@ -141,6 +160,8 @@ fn list_in(cat: &[Entry], root: &Path) -> Vec<ModuleView> {
let version = if e.id == "playerbots" { crate::squid::release(root).map(|release| {
[release.tag, release.commit.map(|commit| commit.chars().take(8).collect())].into_iter().flatten().collect::<Vec<_>>().join(" · ")
}) } else { None };
let key = match e.id.as_str() { "companions" => "bots", "playerbots" => "squid", "content-scaling" => "scaling", "custom-races" => "races", id => id };
let version = if installed { version.or_else(|| installed_commits.get(key).cloned()) } else { None };
ModuleView {
version,
compatibility: if blocked { "unsupported" } else if wildcard && e.id != "client-compat" { "experimental" } else { "compatible" }.into(),
@@ -350,6 +371,12 @@ fn save_settings_in(cat: &[Entry], root: &Path, meta: &Path, id: &str, changes:
return Err(Error::Invalid("Enemy damage must be a number from 0.25 to 2.0.".into()));
}
}
if id == "content-scaling" && key.starts_with("CoAContentScaling.Difficulty.Health.") && key.ends_with("Multiplier") {
let multiplier = value.parse::<f32>().map_err(|_| Error::Invalid("Health multipliers must be a number from 0.1 to 10.".into()))?;
if !multiplier.is_finite() || !(0.1..=10.0).contains(&multiplier) {
return Err(Error::Invalid("Health multipliers must be a number from 0.1 to 10.".into()));
}
}
if id == "content-scaling" && key == "CoAContentScaling.World.Leech.Percent" {
let percent = value.parse::<f32>().map_err(|_| Error::Invalid("Life steal must be a number from 0 to 100 percent.".into()))?;
if !percent.is_finite() || !(0.0..=100.0).contains(&percent) {
@@ -425,6 +452,22 @@ mod tests {
(root, meta)
}
#[test]
fn module_revisions_come_from_the_installed_release_only() {
let temp = tempfile::tempdir().unwrap();
let root = temp.path().join("server");
fs::create_dir_all(&root).unwrap();
let mut meta = crate::registry::InstallMeta::new(crate::registry::InstallKind::Imported, &root);
meta.core.version = Some("2.0.0".into());
let directory = crate::registry::MetaDir::create(&root, &meta).unwrap();
let value = serde_json::json!({"version": "2.0.0", "compatibility": {"moduleCommits": {"races": "a".repeat(40), "bots": "b".repeat(40)}}});
fsx::atomic_write_json(&directory.root.join("manifests/update-2.0.0.json"), &value).unwrap();
assert_eq!(installed_module_commits(&root).get("races").map(String::as_str), Some("aaaaaaaa"));
let stale = serde_json::json!({"version": "1.0.0", "compatibility": {"moduleCommits": {"races": "c".repeat(40)}}});
fsx::atomic_write_json(&directory.root.join("manifests/update-2.0.0.json"), &stale).unwrap();
assert!(installed_module_commits(&root).is_empty());
}
#[test]
fn squid_missing_keys_use_code_defaults_and_saves_enforce_json_ranges() {
let d = tempfile::tempdir().unwrap();
@@ -660,6 +703,36 @@ mod tests {
assert_eq!(settings_in(&cat, &root, "content-scaling").unwrap().iter().find(|s| s.key == key).unwrap().value, "2");
}
#[test]
fn creature_health_multipliers_are_validated_per_key() {
let d = tempfile::tempdir().unwrap();
let (root, meta) = server(d.path());
let cat = catalog();
let conf = dir(&root).join("mod-coa-content-scaling.conf.dist");
let keys = [
"CoAContentScaling.Difficulty.Health.WorldMultiplier",
"CoAContentScaling.Difficulty.Health.DungeonMultiplier",
"CoAContentScaling.Difficulty.Health.RaidMultiplier",
"CoAContentScaling.Difficulty.Health.TrashMultiplier",
"CoAContentScaling.Difficulty.Health.BossMultiplier",
];
let mut text = String::from("CoAContentScaling.Enable = 1\n");
for key in keys {
text.push_str(&format!("{key} = 1.0\n"));
}
fs::write(&conf, text).unwrap();
for key in keys {
let changes = |v: &str| BTreeMap::from([(key.to_string(), v.to_string())]);
for bad in ["NaN", "inf", "0", "0.09", "10.01", "-1", "abc", ""] {
assert!(save_settings_in(&cat, &root, &meta, "content-scaling", &changes(bad)).is_err(), "{key} {bad}");
}
for good in ["0.1", "0.4", "1", "2.5", "10"] {
assert_eq!(save_settings_in(&cat, &root, &meta, "content-scaling", &changes(good)).unwrap(), [key]);
assert_eq!(settings_in(&cat, &root, "content-scaling").unwrap().iter().find(|s| s.key == key).unwrap().value, good);
}
}
}
#[test]
fn an_older_file_name_is_used_when_the_current_one_is_missing_and_a_part_that_cannot_be_switched_stays_on() {
let d = tempfile::tempdir().unwrap();
+242 -125
View File
@@ -153,15 +153,33 @@ pub(crate) fn stop(root: &Path) -> Result<()> {
let observed = crate::process::observe(root, &crate::layout::read_ports(root));
if [observed.mysql.state, observed.auth.state, observed.world.state].iter().all(|state| *state == crate::process::ServiceState::Stopped)
&& !crate::multiworld::is_running(root) { return Ok(()); }
let outcome = crate::driver::run(root, crate::driver::Verb::StopAll)?;
let outcome = crate::driver::run(root, crate::driver::Verb::StopAll)
.map_err(|error| Error::Invalid(format!("Could not stop server {} before copying or restoring MySQL: {error}", root.display())))?;
let status = crate::process::observe(root, &crate::layout::read_ports(root));
if !outcome.ok || [status.mysql.state, status.auth.state, status.world.state].iter().any(|state| *state != crate::process::ServiceState::Stopped)
|| crate::multiworld::is_running(root) {
return Err(Error::Invalid("The complete server must be stopped before copying or restoring MySQL.".into()));
return Err(Error::Invalid(stop_failure(root, &status, &outcome)));
}
Ok(())
}
fn stop_failure(root: &Path, status: &crate::process::Observed, outcome: &crate::driver::DriverOutcome) -> String {
let mut details = vec![format!("The complete server must be stopped before copying or restoring MySQL. Server: {}.", root.display())];
for service in [&status.mysql, &status.auth, &status.world].into_iter().chain(status.secondary_world.iter()) {
if service.state != crate::process::ServiceState::Stopped {
details.push(format!("{}: {:?}; PID: {}; port: {}.", service.name, service.state, service.pid.map(|pid| pid.to_string()).unwrap_or_else(|| "unknown".into()), service.port));
}
if let Some(conflict) = &service.conflict {
details.push(format!("Port {} is used by process {} ({}).", conflict.port, conflict.pid, conflict.exe.as_deref().unwrap_or("executable unavailable")));
}
}
let output = crate::diag::redact(&outcome.output);
let lines: Vec<_> = output.lines().rev().take(12).collect();
if !lines.is_empty() { details.push(lines.into_iter().rev().collect::<Vec<_>>().join("\n")); }
else if !outcome.ok { details.push(format!("The stop command failed (exit code: {:?}).", outcome.exit_code)); }
details.join("\n")
}
pub(crate) fn stop_previous_rehearsals(meta: &Path) -> Result<()> {
let location = meta.join("rehearsals");
if !location.exists() { return Ok(()); }
@@ -226,94 +244,6 @@ pub(crate) fn verify(point: &Path, snapshot: &Snapshot) -> Result<()> {
Ok(())
}
pub(crate) fn rehearsal_copy(root: &Path, meta: &Path, point: &Path, snapshot: &Snapshot) -> Result<PathBuf> {
verify(point, snapshot)?;
let canonical = fsx::canonicalize_lenient(root)?;
let root = canonical.as_path();
let destination = meta.join("rehearsals").join(uuid::Uuid::new_v4().simple().to_string()).join("server");
let mut files = BTreeMap::new();
let mut directories = vec![root.to_path_buf()];
while let Some(directory) = directories.pop() {
for entry in fs::read_dir(directory)? {
let entry = entry?;
let raw = entry.path();
let relative = raw.strip_prefix(root).map_err(|e| Error::Invalid(e.to_string()))?.to_string_lossy().replace('\\', "/");
let lower = relative.to_ascii_lowercase();
let first = lower.split('/').next().unwrap();
if entry.file_type()?.is_dir() && !relative.contains('/')
&& !["core", "data", "scripts", "settings", "runtime", "mysql", ".realms", "bugreport", "extras", "licenses"].contains(&first) { continue; }
if lower == "mysql/data" || lower.starts_with("mysql/data-")
|| lower.split('/').any(|part| part == ".state")
|| lower == "bugreport/reports" || lower == "core/logs" || lower == "mysql/logs" { continue; }
let metadata = fs::symlink_metadata(&raw)?;
reject_link(&metadata)?;
let path = fsx::ensure_within(root, &raw)?;
if metadata.is_dir() { directories.push(path); }
else if metadata.is_file() { files.insert(relative, fsx::sha256_file(&path)?); }
else { return Err(Error::Invalid("Unsupported rehearsal file.".into())); }
}
}
copy_verified(root, &destination, &files)?;
copy_snapshot(point, snapshot, &destination.join("mysql/data"))?;
let (_, mut installed) = crate::registry::MetaDir::open(meta)?;
installed.id = uuid::Uuid::new_v4().to_string();
installed.server_path = destination.to_string_lossy().into_owned();
installed.client_path = None;
crate::registry::MetaDir::create(&destination, &installed)?;
Ok(destination)
}
pub(crate) fn isolate_connections(root: &Path) -> Result<()> {
for relative in ["Settings/worldserver.conf.template", "Settings/authserver.conf.template", "Core/configs/worldserver.conf", "Core/configs/authserver.conf"] {
let path = fsx::ensure_within(root, &root.join(relative))?;
if !path.is_file() {
if relative.ends_with(".template") { return Err(Error::Invalid("The private update copy lacks connection templates.".into())); }
continue;
}
let mut config = crate::config::parser::ConfFile::parse_bytes(&fs::read(&path)?)?;
for key in ["LoginDatabaseInfo", "WorldDatabaseInfo", "CharacterDatabaseInfo"] {
if relative.contains("worldserver") || key == "LoginDatabaseInfo" {
config.set(key, &format!("\"@{key}@\""), &[]);
}
}
config.set("BindIP", "\"127.0.0.1\"", &[]);
if relative.contains("worldserver") {
config.set("Ra.IP", "\"127.0.0.1\"", &[]);
config.set("DataDir", "\"@DATA@\"", &[]);
}
fsx::atomic_write(&path, config.to_text().as_bytes())?;
}
Ok(())
}
pub(crate) fn unchanged(root: &Path, snapshot: &Snapshot) -> Result<()> {
let data = fsx::ensure_within(root, &root.join("mysql/data"))?;
if inventory(&data)? != snapshot.files || directories(&data)? != snapshot.directories {
return Err(Error::Invalid("The installed database changed during the private rehearsal. Retry after stopping other database access.".into()));
}
Ok(())
}
pub(crate) fn validate_connections(templates: &Path, installation: &Path) -> Result<()> {
let port = crate::layout::read_ports(installation).mysql.to_string();
let (_, password) = crate::db::credentials(installation)?;
let realm = crate::realms::state(installation)?.active;
for relative in ["Settings/worldserver.conf.template", "Settings/authserver.conf.template"] {
let path = fsx::ensure_within(templates, &templates.join(relative))?;
let config = crate::config::parser::ConfFile::parse_bytes(&fs::read(path)?)?;
for (key, kind) in [("LoginDatabaseInfo", "auth"), ("WorldDatabaseInfo", "world"), ("CharacterDatabaseInfo", "characters")] {
if relative.contains("authserver") && kind != "auth" { continue; }
let value = config.get(key).ok_or_else(|| Error::Invalid(format!("The update cannot isolate the missing {key} connection.")))?.trim().trim_matches('"');
if value == format!("@{key}@") { continue; }
let fields: Vec<_> = value.split(';').collect();
if fields.len() != 5 || !matches!(fields[0], "127.0.0.1" | "localhost") || fields[1] != port || fields[2] != "acore" || fields[3] != password || fields[4] != realm.schema(kind)? {
return Err(Error::Invalid(format!("The {key} connection does not use this repack's bundled database. A private update cannot safely qualify an external database.")));
}
}
}
Ok(())
}
// Keep the replaced data directory. If interrupted between renames, a retry can install a new verified copy.
pub(crate) fn swap(root: &Path, meta: &Path, point: &Path, snapshot: &Snapshot, fail_after_move: bool) -> Result<PathBuf> {
verify(point, snapshot)?;
@@ -338,6 +268,175 @@ pub(crate) fn swap(root: &Path, meta: &Path, point: &Path, snapshot: &Snapshot,
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn failed_stop_identifies_the_server_process_and_retains_redacted_launcher_reason() {
use crate::process::{Observed, ServiceStatus, ServiceState};
let stopped = |name| ServiceStatus { name, state: ServiceState::Stopped, pid: None, port: 0, port_ready: false, conflict: None, uptime_secs: None };
let status = Observed { mysql: ServiceStatus { state: ServiceState::Stopping, pid: Some(42), port: 3307, ..stopped("mysql") }, auth: stopped("auth"), world: stopped("world"), secondary_world: None };
let outcome = crate::driver::DriverOutcome { ok: false, exit_code: Some(1), code: None, human: None, output: "MySQL is still stopping; wait before copying the repack.\nrootPassword=hunter2".into() };
let message = stop_failure(Path::new("private-rehearsal/server"), &status, &outcome);
assert!(message.contains("private-rehearsal") && message.contains("PID: 42") && message.contains("3307"));
assert!(message.contains("MySQL is still stopping"));
assert!(!message.contains("hunter2"));
}
#[cfg(windows)]
#[test]
#[ignore = "requires COA_MYSQL_TEST_BIN pointing to a repack with Runtime/python and Scripts/manage.py"]
fn real_shutdown_with_broken_templates_preserves_database_rows() {
use std::process::{Child, Command, Stdio};
use std::os::windows::process::CommandExt;
struct Server(Child);
impl Drop for Server { fn drop(&mut self) { let _ = self.0.kill(); let _ = self.0.wait(); } }
let bin = PathBuf::from(std::env::var("COA_MYSQL_TEST_BIN").unwrap());
let source = bin.parent().unwrap().parent().unwrap();
let temp = tempfile::tempdir().unwrap();
let root = temp.path().join("shutdown-fixture");
copy_verified(&bin, &root.join("mysql/bin"), &inventory(&bin).unwrap()).unwrap();
copy_verified(&source.join("Runtime/python"), &root.join("Runtime/python"), &inventory(&source.join("Runtime/python")).unwrap()).unwrap();
fsx::atomic_write(&root.join("Scripts/manage.py"), &fs::read(source.join("Scripts/manage.py")).unwrap()).unwrap();
let data = root.join("mysql/data");
let executable = root.join("mysql/bin/mysqld.exe");
let initialized = Command::new(&executable).args(["--no-defaults", "--initialize-insecure"])
.arg(format!("--datadir={}", data.display())).creation_flags(0x0800_0000).output().unwrap();
assert!(initialized.status.success(), "{}", String::from_utf8_lossy(&initialized.stderr));
let sockets: Vec<_> = (0..4).map(|_| std::net::TcpListener::bind("127.0.0.1:0").unwrap()).collect();
let ports: Vec<_> = sockets.iter().map(|socket| socket.local_addr().unwrap().port()).collect();
fsx::atomic_write_json(&root.join("Settings/repack.json"), &serde_json::json!({"mysqlPort":ports[0],"authPort":ports[1],"worldPort":ports[2],"raPort":ports[3]})).unwrap();
drop(sockets);
fsx::atomic_write_json(&root.join("Settings/database.json"), &serde_json::json!({"rootPassword":"","appPassword":""})).unwrap();
fsx::atomic_write(&root.join("mysql/admin-client.ini"), format!("[client]\nuser=root\nhost=127.0.0.1\nport={}\nprotocol=tcp\n", ports[0]).as_bytes()).unwrap();
fsx::atomic_write(&root.join("Settings/worldserver.conf.template"), b"WorldDatabaseInfo = @UNKNOWN_DATABASE@\n").unwrap();
let original = fs::read(root.join("Settings/worldserver.conf.template")).unwrap();
let db = crate::db::Db::from_repack(&root, crate::db::Account::Admin).unwrap();
let start = || {
let server = Server(Command::new(&executable).args(["--no-defaults", "--bind-address=127.0.0.1", "--mysqlx=0", "--skip-log-bin"])
.arg(format!("--datadir={}", data.display())).arg(format!("--port={}", ports[0]))
.arg(format!("--log-error={}", root.join("mysql-test.log").display()))
.stdout(Stdio::null()).stderr(Stdio::null()).creation_flags(0x0800_0000).spawn().unwrap());
let deadline = std::time::Instant::now() + std::time::Duration::from_secs(60);
while !db.ping() { assert!(std::time::Instant::now() < deadline); std::thread::sleep(std::time::Duration::from_millis(200)); }
fsx::atomic_write_json(&root.join(".state/mysql.json"), &crate::process::identity(server.0.id()).unwrap()).unwrap();
server
};
let mut server = start();
db.query("CREATE DATABASE acore_characters; CREATE TABLE acore_characters.shutdown_probe(id INT PRIMARY KEY); INSERT INTO acore_characters.shutdown_probe VALUES(73);").unwrap();
db.query("CREATE DATABASE acore_auth; CREATE DATABASE acore_world; CREATE VIEW acore_characters.shutdown_view AS SELECT id FROM acore_characters.shutdown_probe;").unwrap();
let copied_configuration = serde_json::json!({"root":temp.path().join("original-installation"),"ports":{"mysqlPort":ports[0],"authPort":ports[1],"worldPort":ports[2],"raPort":ports[3]}});
fsx::atomic_write_json(&root.join(".state/configuration.json"), &copied_configuration).unwrap();
let meta = crate::registry::metadata_dir_for(&root).unwrap();
crate::registry::MetaDir::create(&root, &crate::registry::InstallMeta::new(crate::registry::InstallKind::New, &root)).unwrap();
let point = crate::backup::create(&root, &meta, crate::backup::Kind::Full, crate::backup::Trigger::BeforeRepair, None, &|_| {}).unwrap();
assert!(point.mysql_snapshot.is_some());
assert!(crate::backup::verify(&meta, &point.id).unwrap().ok);
crate::update::RepackEnv { root: &root, meta_dir: &meta }.verify_recovery_point(&point).unwrap();
assert!(server.0.wait().unwrap().success());
assert_eq!(fs::read(root.join("Settings/worldserver.conf.template")).unwrap(), original);
assert_eq!(fsx::read_json::<serde_json::Value>(&root.join(".state/configuration.json")).unwrap(), copied_configuration);
fs::remove_file(root.join(".state/configuration.json")).unwrap();
let mut restarted = start();
assert_eq!(db.query("SELECT id FROM acore_characters.shutdown_probe;").unwrap(), "73");
fsx::atomic_write_json(&root.join(".state/mysql.json"), &crate::process::identity(std::process::id()).unwrap()).unwrap();
assert!(stop(&root).unwrap_err().to_string().contains("different live process"));
assert_eq!(db.query("SELECT id FROM acore_characters.shutdown_probe;").unwrap(), "73");
fs::remove_file(root.join("Settings/worldserver.conf.template")).unwrap();
fs::remove_file(root.join(".state/mysql.json")).unwrap();
let observed = crate::process::observe(&root, &crate::layout::read_ports(&root));
assert_eq!(observed.mysql.state, crate::process::ServiceState::Running);
assert!(observed.mysql.port_ready);
assert!(crate::process::listeners().contains(&(ports[0], observed.mysql.pid.unwrap())));
stop(&root).unwrap();
assert!(restarted.0.wait().unwrap().success());
let mut third = start();
assert_eq!(db.query("SELECT id FROM acore_characters.shutdown_view;").unwrap(), "73");
db.query("UPDATE acore_characters.shutdown_probe SET id=74;").unwrap();
fsx::atomic_write(&root.join(".state/mysql.json"), b"{truncated").unwrap();
stop(&root).unwrap();
assert!(third.0.wait().unwrap().success());
assert_eq!(fs::read(root.join(".state/mysql-before-shutdown-recovery.json")).unwrap(), b"{truncated");
let point_dir = crate::backup::point_json(&meta, &point.id).unwrap().parent().unwrap().to_path_buf();
let retained = swap(&root, &meta, &point_dir, point.mysql_snapshot.as_ref().unwrap(), false).unwrap();
assert!(retained.is_dir());
let mut recovered = start();
assert_eq!(db.query("SELECT id FROM acore_characters.shutdown_view;").unwrap(), "73");
stop(&root).unwrap();
assert!(recovered.0.wait().unwrap().success());
// Exercise the public Repair pipeline with real signed packages and SQL,
// rather than only the snapshot and journal helpers.
let mut package_server = start();
db.query("CREATE TABLE acore_auth.account(id INT PRIMARY KEY, username VARCHAR(40)); INSERT INTO acore_auth.account VALUES(73,'FIXTURE'); CREATE TABLE acore_auth.realmlist(id INT PRIMARY KEY, name VARCHAR(40), address VARCHAR(40), localAddress VARCHAR(40), port INT, flag INT); INSERT INTO acore_auth.realmlist VALUES(1,'AzerothCore','127.0.0.1','127.0.0.1',18085,0); CREATE TABLE acore_characters.characters(guid INT PRIMARY KEY, account INT, name VARCHAR(40), race INT, class INT); INSERT INTO acore_characters.characters VALUES(73,73,'Fixture',7,14); CREATE TABLE acore_world.playercreateinfo(race INT, class INT, PRIMARY KEY(race,class));").unwrap();
let starts = [1,2,3,4,5,6,7,8,10,11].into_iter().flat_map(|race| (12..=32).map(move |class| format!("({race},{class})"))).collect::<Vec<_>>().join(",");
db.query(&format!("INSERT INTO acore_world.playercreateinfo VALUES {starts};")).unwrap();
let password = "a".repeat(48);
db.query(&format!("ALTER USER 'root'@'localhost' IDENTIFIED BY '{password}';")).unwrap();
fsx::atomic_write_json(&root.join("Settings/database.json"), &serde_json::json!({"rootPassword":password,"appPassword":"b".repeat(48)})).unwrap();
fsx::atomic_write(&root.join("Settings/worldserver.conf.template"), b"WorldDatabaseInfo = @WorldDatabaseInfo@\nCharacterDatabaseInfo = @CharacterDatabaseInfo@\nLoginDatabaseInfo = @LoginDatabaseInfo@\n").unwrap();
fsx::atomic_write(&root.join("Settings/authserver.conf.template"), b"LoginDatabaseInfo = @LoginDatabaseInfo@\n").unwrap();
fsx::atomic_write(&root.join("Settings/mod_ascension_compat.conf.template"), b"Ascension.DbcPath = @ASCENSION_DBC@\n").unwrap();
let startup = crate::driver::run(&root, crate::driver::Verb::StartMysql).unwrap();
assert!(startup.ok, "{}", crate::diag::redact(&startup.output));
let db = crate::db::Db::from_repack(&root, crate::db::Account::Admin).unwrap();
crate::schema_check::capture(&db, &root).unwrap();
assert!(crate::schema_check::check(&db, &root).unwrap().is_empty());
stop(&root).unwrap();
assert!(package_server.0.wait().unwrap().success());
struct StopFixture(PathBuf);
impl Drop for StopFixture { fn drop(&mut self) { let _ = stop(&self.0); } }
let _stop_fixture = StopFixture(root.clone());
use base64::{engine::general_purpose::STANDARD, Engine};
use ed25519_dalek::{Signer, SigningKey};
let key = SigningKey::from_bytes(&[7;32]);
let trusted = STANDARD.encode(key.verifying_key().to_bytes());
let source = temp.path().join("repair-package-source");
fsx::atomic_write(&source.join("Core/reference/repair-probe.dll"), b"official").unwrap();
fsx::atomic_write(&source.join(crate::schema_check::CONTRACT), &fs::read(root.join(crate::schema_check::CONTRACT)).unwrap()).unwrap();
let build = |kind, version: &str, destination: &Path, migrations| {
let manifest = crate::package::build(&source, destination, &crate::package::BuildOptions { kind, version: version.into(), core_commit: None, built_at: "fixture".into(), part_size: 1024*1024, bots_commit: None, migrations }, &|_| {}).unwrap();
let bytes = fs::read(destination.join("manifest.json")).unwrap();
fsx::atomic_write(&destination.join("manifest.json.sig"), STANDARD.encode(key.sign(&bytes).to_bytes()).as_bytes()).unwrap();
manifest
};
let base = temp.path().join("repair-base");
let base_manifest = build(crate::manifest::Kind::Base, "1.0.0", &base, vec![]);
let update = temp.path().join("repair-update");
let update_manifest = build(crate::manifest::Kind::Update, "2.0.0", &update, vec![]);
fsx::atomic_write_json(&meta.join("manifests/base.json"), &base_manifest).unwrap();
fsx::atomic_write_json(&meta.join("manifests/update-2.0.0.json"), &update_manifest).unwrap();
let (_, mut installed) = crate::registry::MetaDir::open(&meta).unwrap();
installed.core.version = Some("2.0.0".into());
fsx::atomic_write_json(&meta.join("install.json"), &installed).unwrap();
fsx::atomic_write(&root.join("Core/reference/repair-probe.dll"), b"broken").unwrap();
let base_source = crate::pkgsource::Source::Dir(base);
let update_source = crate::pkgsource::Source::Dir(update);
let report = crate::repair::run(&root, &meta, &base_source, &update_source, &trusted, &|_,_| {}).unwrap();
assert!(report.error.is_none());
assert!(report.database.iter().all(|check| check.problems.is_empty()));
assert_eq!(fs::read(root.join("Core/reference/repair-probe.dll")).unwrap(), b"official");
crate::update::ensure_recovered(&meta).unwrap();
fsx::atomic_write(&root.join("Core/reference/repair-probe.dll"), b"before verification failure").unwrap();
let error = crate::repair::run(&root, &meta, &base_source, &update_source, &trusted, &|phase,_| {
if phase == "Checking database after repair" { fs::remove_file(root.join("Core/reference/repair-probe.dll")).unwrap(); }
}).unwrap_err();
assert!(error.to_string().contains("repair"));
assert_eq!(fs::read(root.join("Core/reference/repair-probe.dll")).unwrap(), b"before verification failure");
crate::update::ensure_recovered(&meta).unwrap();
let failed_sql = "UPDATE shutdown_probe SET id=75; CREATE TABLE repair_partial(id INT); INVALID SQL;";
fsx::atomic_write(&source.join("_migrations/characters/repair-failure.sql"), failed_sql.as_bytes()).unwrap();
let failing = temp.path().join("repair-failing-update");
let manifest = build(crate::manifest::Kind::Update, "2.0.0", &failing, vec![crate::manifest::Migration { id:"repair-failure".into(), db:"characters".into(), sha256:fsx::sha256_bytes(failed_sql.as_bytes()), compatible_sha256:vec![], destructive:false }]);
fsx::atomic_write_json(&meta.join("manifests/update-2.0.0.json"), &manifest).unwrap();
fsx::atomic_write(&root.join("Core/reference/repair-probe.dll"), b"before failed repair").unwrap();
let report = crate::repair::run(&root, &meta, &base_source, &crate::pkgsource::Source::Dir(failing), &trusted, &|_,_| {}).unwrap();
assert!(report.error.is_some());
assert_eq!(fs::read(root.join("Core/reference/repair-probe.dll")).unwrap(), b"before failed repair");
crate::update::ensure_recovered(&meta).unwrap();
crate::backup::with_database(&root, |db| {
assert_eq!(db.query("SELECT id FROM acore_characters.shutdown_view;")?, "73");
assert_eq!(db.query("SELECT COUNT(*) FROM information_schema.TABLES WHERE TABLE_SCHEMA='acore_characters' AND TABLE_NAME='repair_partial';")?, "0");
assert_eq!(db.query("SELECT username FROM acore_auth.account WHERE id=73;")?, "FIXTURE");
Ok(())
}).unwrap();
}
#[cfg(windows)]
#[test]
#[ignore = "requires COA_MYSQL_TEST_BIN pointing to the bundled MySQL tools"]
@@ -372,7 +471,55 @@ mod tests {
};
let mut server = start();
db.query("CREATE DATABASE acore_characters; CREATE DATABASE acore_world; CREATE DATABASE acore_auth; CREATE TABLE acore_characters.probe(id INT PRIMARY KEY, note VARCHAR(40)); CREATE TABLE acore_world.probe(id INT NOT NULL DEFAULT 0); INSERT INTO acore_world.probe VALUES(43); CREATE TABLE acore_auth.probe(id INT); INSERT INTO acore_auth.probe VALUES(44); CREATE USER 'fixture_user'@'localhost' IDENTIFIED BY 'fixture'; CREATE TRIGGER acore_characters.probe_insert BEFORE INSERT ON acore_characters.probe FOR EACH ROW SET NEW.note='triggered'; INSERT INTO acore_characters.probe VALUES(42,'original'); CREATE VIEW acore_characters.probe_view AS SELECT c.id,w.id AS world_id FROM acore_characters.probe c CROSS JOIN acore_world.probe w; CREATE PROCEDURE acore_characters.probe_proc() SELECT id FROM acore_characters.probe; CREATE FUNCTION acore_characters.probe_func() RETURNS INT DETERMINISTIC RETURN 73;").unwrap();
// Shared auth must not run again when the active realm changes.
use crate::migrations::Store;
db.query("CREATE DATABASE acore_world_wildcard;").unwrap();
let migration_dir = root.join("migration-fixture");
fs::create_dir_all(migration_dir.join("auth")).unwrap();
fs::create_dir_all(migration_dir.join("world")).unwrap();
let auth_sql = b"INSERT INTO acore_auth.probe VALUES(45);";
fs::write(migration_dir.join("auth/shared_once.sql"), auth_sql).unwrap();
let auth_list = vec![crate::manifest::Migration { id:"shared_once".into(), db:"auth".into(), sha256:fsx::sha256_bytes(auth_sql), destructive:false, compatible_sha256:vec![] }];
assert_eq!(crate::migrations::apply_pending(&db, &auth_list, &migration_dir, &|| Ok("fixture".into())).unwrap().applied.len(), 1);
let wildcard = db.clone().for_realm(crate::realms::Mode::Wildcard);
assert_eq!(crate::migrations::pending_count(&wildcard, &auth_list).unwrap(), 0, "auth is shared by both realms");
assert!(crate::migrations::apply_pending(&wildcard, &auth_list, &migration_dir, &|| Ok("fixture".into())).unwrap().applied.is_empty());
assert_eq!(db.query("SELECT COUNT(*) FROM acore_auth.probe WHERE id=45;").unwrap(), "1");
let mut failed = crate::migrations::LedgerRow { db:"auth".into(), id:"shared_once".into(), sha256:auth_list[0].sha256.clone(), status:crate::migrations::Status::Failed, error:Some("interrupted fixture".into()), baseline:false };
db.put(&failed).unwrap();
assert!(crate::migrations::pending_count(&wildcard, &auth_list).is_err());
failed.status = crate::migrations::Status::Applied; failed.error = None; db.put(&failed).unwrap();
// Legacy per-world auth records are adopted without replaying their SQL.
let legacy_sql = b"INSERT INTO acore_auth.probe VALUES(46);";
fs::write(migration_dir.join("auth/legacy_auth.sql"), legacy_sql).unwrap();
db.query(&format!("INSERT INTO acore_auth.probe VALUES(46); REPLACE INTO acore_world.coa_manager_migrations VALUES('auth','legacy_auth','{}','applied',NOW(),NULL,0);", fsx::sha256_bytes(legacy_sql))).unwrap();
let legacy_auth = vec![crate::manifest::Migration { id:"legacy_auth".into(), db:"auth".into(), sha256:fsx::sha256_bytes(legacy_sql), destructive:false, compatible_sha256:vec![] }];
assert_eq!(crate::migrations::pending_count(&wildcard, &legacy_auth).unwrap(), 0);
assert!(crate::migrations::apply_pending(&wildcard, &legacy_auth, &migration_dir, &|| Ok("fixture".into())).unwrap().applied.is_empty());
assert_eq!(db.query("SELECT COUNT(*) FROM acore_auth.probe WHERE id=46;").unwrap(), "1");
// A core history cleanup cannot make previously recorded SQL pending again.
db.query("CREATE TABLE acore_world.updates(name VARCHAR(150) PRIMARY KEY) DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_unicode_ci; INSERT INTO acore_world.updates VALUES('LEGACY_WORLD.sql'); INSERT INTO acore_world.probe VALUES(99);").unwrap();
db.load().unwrap();
db.query("DROP TABLE acore_world.updates;").unwrap();
let legacy_world_sql = b"INSERT INTO acore_world.probe VALUES(99);";
fs::write(migration_dir.join("world/legacy_world.sql"), legacy_world_sql).unwrap();
let legacy_world = vec![crate::manifest::Migration { id:"legacy_world".into(), db:"world".into(), sha256:fsx::sha256_bytes(legacy_world_sql), destructive:false, compatible_sha256:vec![] }];
assert_eq!(crate::migrations::pending_count(&db, &legacy_world).unwrap(), 0);
assert!(crate::migrations::apply_pending(&db, &legacy_world, &migration_dir, &|| Ok("fixture".into())).unwrap().applied.is_empty());
assert_eq!(db.query("SELECT COUNT(*) FROM acore_world.probe WHERE id=99;").unwrap(), "1");
db.query("DELETE FROM acore_world.probe WHERE id=99;").unwrap();
assert_eq!(db.recovery_objects("acore_characters").unwrap().len(), 4);
db.query("CREATE TABLE acore_characters.coa_squid_pending(path VARCHAR(255), sha256 CHAR(64)); INSERT INTO acore_characters.coa_squid_pending VALUES('playerbots/base/ai_playerbot_texts.sql', REPEAT('a',64));").unwrap();
let squid_problems = crate::squid::database_problems(&root, crate::realms::Mode::Coa, &|sql| {
assert!(sql.starts_with("SELECT "));
db.query(sql)
}).unwrap();
assert_eq!(squid_problems.len(), 1);
assert!(squid_problems[0].detail.contains("interrupted"));
assert_eq!(db.query("SELECT COUNT(*) FROM acore_characters.coa_squid_pending;").unwrap(), "1");
crate::schema_check::capture(&db, &root).unwrap();
let hash = fsx::sha256_file(&root.join(crate::schema_check::CONTRACT)).unwrap();
db.query("ALTER TABLE acore_world.probe ALTER COLUMN id DROP DEFAULT;").unwrap();
@@ -454,40 +601,10 @@ mod tests {
fsx::atomic_write(&archive_path, b"damaged").unwrap();
assert!(verify(&point, &snapshot).is_err());
}
#[test]
fn private_rehearsals_reject_external_database_connections() {
let (_temp, root, _meta, _point, _snapshot) = fixture();
fsx::atomic_write_json(&root.join("Settings/repack.json"), &serde_json::json!({"mysqlPort":3307})).unwrap();
fsx::atomic_write_json(&root.join("Settings/database.json"), &serde_json::json!({"rootPassword":"test-root","appPassword":"test-app"})).unwrap();
fsx::atomic_write(&root.join("Settings/worldserver.conf.template"), b"LoginDatabaseInfo = \"@LoginDatabaseInfo@\"\nWorldDatabaseInfo = \"@WorldDatabaseInfo@\"\nCharacterDatabaseInfo = \"@CharacterDatabaseInfo@\"\n").unwrap();
fsx::atomic_write(&root.join("Settings/authserver.conf.template"), b"LoginDatabaseInfo = \"127.0.0.1;3307;acore;test-app;acore_auth\"\n").unwrap();
validate_connections(&root, &root).unwrap();
fsx::atomic_write(&root.join("Settings/authserver.conf.template"), b"LoginDatabaseInfo = \"192.0.2.1;3307;acore;test-app;acore_auth\"\n").unwrap();
let error = validate_connections(&root, &root).unwrap_err().to_string();
assert!(error.contains("external database"));
assert!(!error.contains("test-app"));
}
#[test]
fn a_changed_live_database_invalidates_the_rehearsal() {
let (_temp, root, _meta, _point, snapshot) = fixture();
assert!(unchanged(&root, &snapshot).is_err());
fsx::atomic_write(&root.join("mysql/data/characters.ibd"), b"saved").unwrap();
fs::create_dir_all(root.join("mysql/data/empty_schema")).unwrap();
unchanged(&root, &snapshot).unwrap();
fsx::atomic_write(&root.join("mysql/data/extra.ibd"), b"new data").unwrap();
assert!(unchanged(&root, &snapshot).is_err());
}
#[test]
fn rehearsal_copy_succeeds_with_nested_directories_and_unnormalized_root() {
let (_temp, root, meta, point, snapshot) = fixture();
let meta_obj = crate::registry::InstallMeta::new(crate::registry::InstallKind::New, &root);
fsx::atomic_write_json(&meta.join("install.json"), &meta_obj).unwrap();
fsx::atomic_write(&root.join("Core/worldserver.exe"), b"exe").unwrap();
fsx::atomic_write(&root.join("Core/configs/worldserver.conf"), b"conf").unwrap();
fsx::atomic_write(&root.join("Settings/repack.json"), b"{}").unwrap();
let dest = rehearsal_copy(&root, &meta, &point, &snapshot).unwrap();
assert_eq!(fs::read(dest.join("Core/configs/worldserver.conf")).unwrap(), b"conf");
assert_eq!(fs::read(dest.join("mysql/data/characters.ibd")).unwrap(), b"saved");
}
}
+43
View File
@@ -14,6 +14,7 @@ use crate::signing;
/// Where a package comes from.
#[derive(Debug, Clone)]
pub enum Source {
Installation(Box<Source>),
/// `https://.../` base URL holding manifest.json, manifest.json.sig and the parts.
Url(String),
/// A folder produced by `package::build` plus its `manifest.json.sig` (offline / local installs).
@@ -23,6 +24,7 @@ pub enum Source {
/// Read a small file (manifest, signature) from the source.
pub fn fetch_small(source: &Source, name: &str) -> Result<Vec<u8>> {
match source {
Source::Installation(source) => fetch_small(source, &if matches!(name, "manifest.json" | "manifest.json.sig") { format!("install-{name}") } else { name.into() }),
Source::Dir(d) => Ok(fs::read(fsx::safe_join(d, name)?).map_err(|_| Error::Invalid(format!("{name} was not found in the package folder.")))?),
Source::Url(base) => {
let url = format!("{}/{name}", base.trim_end_matches('/'));
@@ -48,6 +50,14 @@ pub fn fetch_manifest(source: &Source, trusted_key: &str) -> Result<(Manifest, V
let sig = String::from_utf8_lossy(&fetch_small(source, "manifest.json.sig")?).into_owned();
signing::verify(&bytes, &sig, trusted_key)?;
let manifest = Manifest::parse(&bytes)?;
if let Source::Installation(release) = source {
let (update, _) = fetch_manifest(release, trusted_key)?;
if manifest.kind != crate::manifest::Kind::Base || update.kind != crate::manifest::Kind::Update || manifest.version != update.version
|| manifest.core.commit != update.core.commit
|| manifest.bots.as_ref().and_then(|value| value.commit.as_ref()) != update.bots.as_ref().and_then(|value| value.commit.as_ref()) {
return Err(Error::InvalidManifest("The full installation package does not match the selected server release.".into()));
}
}
Ok((manifest, bytes))
}
@@ -68,6 +78,7 @@ pub fn human_bytes(n: u64) -> String {
pub fn fetch_parts(source: &Source, manifest: &Manifest, download_dir: &Path, cancel: &Cancel, report: &dyn Fn(f64, Option<String>)) -> Result<PathBuf> {
let archive = manifest.archive.as_ref().ok_or_else(|| Error::InvalidManifest("manifest has no archive".into()))?;
match source {
Source::Installation(source) => fetch_parts(source, manifest, download_dir, cancel, report),
Source::Dir(d) => Ok(d.clone()),
Source::Url(base) => {
let all: u64 = archive.parts.iter().map(|p| p.size).sum::<u64>().max(1);
@@ -90,6 +101,38 @@ pub fn fetch_parts(source: &Source, manifest: &Manifest, download_dir: &Path, ca
mod tests {
use super::*;
#[test]
fn clean_installation_requires_the_signed_full_image_of_the_selected_release() {
use base64::{engine::general_purpose::STANDARD, Engine};
use ed25519_dalek::{Signer, SigningKey};
let folder = tempfile::tempdir().unwrap();
let signing_key = SigningKey::from_bytes(&[23; 32]);
let key = STANDARD.encode(signing_key.verifying_key().to_bytes());
let source = Source::Installation(Box::new(Source::Dir(folder.path().into())));
let mut manifest = serde_json::json!({"schema": 1, "kind": "update", "version": "0.261014.1",
"core": {"commit": "a".repeat(40)}, "builtAt": "test", "minManagerVersion": "0.6.16",
"files": [], "migrations": [], "archive": null});
let write = |name: &str, value: &serde_json::Value| {
let bytes = serde_json::to_vec(value).unwrap();
fs::write(folder.path().join(name), &bytes).unwrap();
fs::write(folder.path().join(format!("{name}.sig")), STANDARD.encode(signing_key.sign(&bytes).to_bytes())).unwrap();
};
write("manifest.json", &manifest);
assert!(fetch_manifest(&source, &key).is_err());
manifest["kind"] = "base".into();
write("install-manifest.json", &manifest);
assert_eq!(fetch_manifest(&source, &key).unwrap().0.version, "0.261014.1");
manifest["version"] = "0.2.0".into();
write("install-manifest.json", &manifest);
assert!(fetch_manifest(&source, &key).is_err());
manifest["version"] = "0.261014.1".into();
manifest["core"]["commit"] = "b".repeat(40).into();
write("install-manifest.json", &manifest);
assert!(fetch_manifest(&source, &key).is_err());
fs::write(folder.path().join("install-manifest.json"), b"tampered").unwrap();
assert!(fetch_manifest(&source, &key).unwrap_err().to_string().contains("signature"));
}
#[test]
fn sizes_are_shown_in_binary_units() {
assert_eq!(human_bytes(0), "0 B");
+19 -1
View File
@@ -71,7 +71,15 @@ pub fn hardware() -> Hardware {
#[cfg(not(windows))]
pub fn hardware() -> Hardware {
Hardware { cores: std::thread::available_parallelism().map(|n| n.get() as u32).unwrap_or(2), ram_gb: 8.0, free_ram_gb: 4.0 }
let (total, free) = std::fs::read_to_string("/proc/meminfo").ok().and_then(|t| parse_meminfo(&t)).unwrap_or((8.0, 4.0));
Hardware { cores: std::thread::available_parallelism().map(|n| n.get() as u32).unwrap_or(2), ram_gb: total, free_ram_gb: free }
}
/// (total, available) memory in GB from the text of `/proc/meminfo`.
#[cfg(not(windows))]
fn parse_meminfo(text: &str) -> Option<(f64, f64)> {
let kb = |key: &str| -> Option<f64> { text.lines().find_map(|l| l.strip_prefix(key)?.trim().strip_suffix("kB")?.trim().parse::<f64>().ok()) };
Some((kb("MemTotal:")? / (1u64 << 20) as f64, kb("MemAvailable:")? / (1u64 << 20) as f64))
}
#[derive(Debug, Clone, Serialize)]
@@ -105,6 +113,16 @@ pub fn sizes(h: &Hardware) -> Vec<SizeOption> {
mod tests {
use super::*;
#[cfg(not(windows))]
#[test]
fn memory_is_read_from_proc_meminfo() {
let t = "MemTotal: 65536000 kB\nMemFree: 1000000 kB\nMemAvailable: 16777216 kB\nBuffers: 1 kB\n";
let (total, free) = parse_meminfo(t).unwrap();
assert!((total - 62.5).abs() < 0.01 && (free - 16.0).abs() < 0.01, "{total} {free}");
assert!(parse_meminfo("nothing useful").is_none());
assert!(hardware().ram_gb > 0.0);
}
#[test]
fn parses_counts_and_never_goes_negative() {
let p = parse_counts("12\t9\t400").unwrap();
+9 -6
View File
@@ -289,15 +289,18 @@ fn service(root: &Path, name: &'static str, exe: PathBuf, port: u16, listen: &[(
}
fn service_with(root: &Path, name: &'static str, exe: PathBuf, alternatives: &[PathBuf], port: u16, listen: &[(u16, u32)]) -> ServiceStatus {
// Prefer the repack's own record when it still matches a live process; otherwise discover by exact exe path.
// Prefer the exact installed executable owning the port; otherwise use a valid record or path discovery.
let norm = |p: &Path| p.to_string_lossy().replace('/', "\\");
let expected: Vec<String> = std::iter::once(&exe).chain(alternatives).map(|p| norm(p)).collect();
let proc = read_state_record(root, name)
.filter(is_alive)
.filter(|r| expected.iter().any(|e| r.exe.eq_ignore_ascii_case(e)))
.or_else(|| std::iter::once(&exe).chain(alternatives).find_map(|p| find_by_exe(p).into_iter().next()));
let owner = listen.iter().find(|(p, _)| *p == port).map(|(_, pid)| *pid);
// MySQL can launch a supervisor and a server with the same executable. The
// listening process is authoritative when its exact installation path matches.
let proc = owner.and_then(identity)
.filter(|record| expected.iter().any(|exe| record.exe.eq_ignore_ascii_case(exe)))
.or_else(|| read_state_record(root, name)
.filter(is_alive)
.filter(|r| expected.iter().any(|e| r.exe.eq_ignore_ascii_case(e))))
.or_else(|| std::iter::once(&exe).chain(alternatives).find_map(|p| find_by_exe(p).into_iter().next()));
let (state, port_ready, conflict) = match (&proc, owner) {
(Some(p), Some(o)) if p.pid == o => (ServiceState::Running, true, None),
(Some(_), _) => (ServiceState::Starting, false, None),
+2 -1
View File
@@ -348,7 +348,8 @@ AC>";
let (port, _h) = fake_ra("Account not found.");
let mut ra = Ra::connect_to(port, "u", "p").unwrap();
assert!(ra.set_account_password("x", "newpass9").is_err(), "the name is validated before anything is sent");
assert!(ra.set_account_password("Nobody", "newpass9").unwrap_err().to_string().contains("not found"));
let error = ra.set_account_password("Nobody", "newpass9").unwrap_err().to_string();
assert!(error.contains("not found"), "Unexpected console error: {error}");
}
#[test]
+101 -3
View File
@@ -1,5 +1,6 @@
//! Database acceptance for release packages, using a freshly extracted signed base fixture.
use std::collections::HashSet;
use std::fs;
use std::path::Path;
use crate::{backup, fsx, install, manifest::{Kind, Migration}, migrations, package, pkgsource, release::{self, SqlFile}, schema_check, Error, Result};
@@ -13,7 +14,6 @@ impl crate::update::Env for QualificationEnv<'_> {
fn preflight(&self, manifest: &crate::manifest::Manifest) -> Result<()> { self.0.preflight(manifest) }
fn snapshot(&self) -> Result<String> { self.0.snapshot() }
fn verify_snapshot(&self, id: &str) -> Result<()> { self.0.verify_snapshot(id) }
fn rehearse(&self, manifest: &crate::manifest::Manifest, tree: &Path, items: &[crate::update::PlanItem], point: &str) -> Result<()> { self.0.rehearse(manifest, tree, items, point) }
fn restore_snapshot(&self, id: &str) -> Result<()> { self.0.restore_snapshot(id) }
fn migrate(&self, manifest: &crate::manifest::Manifest, staged: &Path) -> Result<crate::migrations::ApplyReport> { self.0.migrate(manifest, staged) }
fn validate(&self) -> Result<()> { self.0.validate() }
@@ -85,10 +85,72 @@ pub fn qualify_upgrade(base: &Path, source: &Path, candidate: &Path, fixture: &P
"schema": 1, "result": "passed", "fromVersion": source_manifest.version,
"toVersion": candidate_manifest.version, "fromManifestSha256": fsx::sha256_bytes(&source_bytes),
"fromSchemaSha256": source_manifest.files.iter().find(|file| file.path == schema_check::CONTRACT).map(|file| &file.sha256),
"checks": ["manager-transaction", "private-user-database-rehearsal", "database-schema", "private-auth-world-startup", "existing-identities", "user-file-preservation"],
"checks": ["manager-transaction", "recovery-backed-database-update", "database-schema", "private-auth-world-startup", "existing-identities", "user-file-preservation"],
}))
}
pub fn compose_install(base: &Path, candidate: &Path, fixture: &Path, trusted_key: &str) -> Result<()> {
qualify_upgrade(base, base, candidate, fixture, trusted_key)?;
let fixture = fsx::canonicalize_lenient(fixture)?;
let (_, bytes) = pkgsource::fetch_manifest(&pkgsource::Source::Dir(candidate.into()), trusted_key)?;
let manifest = crate::manifest::Manifest::parse(&bytes)?;
install::with_scratch_ports(&fixture, || backup::with_database(&fixture, |db| {
let players = db.query(&format!("SELECT COUNT(*) FROM acore_characters.characters WHERE account<>{};", crate::companions::TEMPLATE_ACCOUNT))?;
if players.trim() != "0" || crate::companions::template_count(db)? != 21 {
return Err(Error::Invalid("The installation image contains unexpected player characters.".into()));
}
for table in crate::cleanbase::AUTH_PLAYER_TABLES {
if db.tables("acore_auth")?.iter().any(|name| name == table) {
db.query(&format!("DELETE FROM acore_auth.`{table}`;"))?;
}
}
if let Some(problem) = schema_check::check(db, &fixture)?.first() {
return Err(Error::Invalid(format!("Installation image schema mismatch: {}.{}", problem.table, problem.column)));
}
Ok(())
}))?;
fs::remove_file(fixture.join("Core/configs/coa-qualification-user.conf"))?;
let path = fixture.join("Settings/repack.json");
let mut settings: serde_json::Value = fsx::read_json(&path)?;
settings["raPassword"] = "".into();
fsx::atomic_write_json(&path, &settings)?;
fsx::atomic_write_json(&fixture.join("Settings/install-source.json"), &serde_json::json!({
"version": manifest.version, "core": manifest.core, "bots": manifest.bots,
"updateManifestSha256": fsx::sha256_bytes(&bytes),
}))?;
crate::mysql_snapshot::stop(&fixture)?;
fs::remove_file(fixture.join(MARKER))?;
Ok(())
}
pub fn qualify_install(package: &Path, fixture: &Path, trusted_key: &str) -> Result<serde_json::Value> {
use crate::update::Env;
if !cfg!(windows) || fixture.exists()
|| !fixture.file_name().is_some_and(|name| name.to_string_lossy().starts_with("coa-schema-fixture-")) {
return Err(Error::Invalid("Clean installation qualification requires a new disposable Windows fixture.".into()));
}
let fixture = std::path::absolute(fixture)?;
let package = fsx::canonicalize_lenient(package)?;
let (manifest, bytes) = pkgsource::fetch_manifest(&pkgsource::Source::Dir(package.clone()), trusted_key)?;
let registry_dir = tempfile::tempdir()?;
let registry = crate::registry::Registry::at(registry_dir.path().join("registry.json"));
let installed = install::install_base(&install::Params {
source: pkgsource::Source::Dir(package), dest: fixture.clone(), trusted_key,
registry: &registry, cancel: crate::download::Cancel::default(),
}, &|_| {})?;
let metadata = crate::registry::metadata_dir_for(&fixture)?;
let env = QualificationEnv(crate::update::RepackEnv { root: &fixture, meta_dir: &metadata });
env.validate()?;
env.ensure_stopped()?;
crate::mysql_snapshot::stop(&fixture)?;
if installed.version != manifest.version || crate::update::pending_checked(&metadata)?.is_some() {
return Err(Error::Invalid("Clean installation did not finish at the packaged release.".into()));
}
Ok(serde_json::json!({ "schema": 1, "result": "passed", "version": installed.version,
"manifestSha256": fsx::sha256_bytes(&bytes),
"checks": ["clean-install", "signed-full-package", "credential-rotation", "schema-contract", "private-auth-world-startup"] }))
}
pub fn collect(core: &Path, bots: Option<&Path>, repairs: Option<&Path>) -> Result<Vec<SqlFile>> {
let mut files = release::collect_core_sql(core)?;
if let Some(bots) = bots { files.extend(release::collect_bots_sql(bots)?); }
@@ -220,7 +282,15 @@ pub fn validate_startup(fixture: &Path, tree: &Path, base: &Path, trusted_key: &
}
pub fn verify_package(dir: &Path, trusted_key: &str) -> Result<()> {
let (manifest, _) = pkgsource::fetch_manifest(&pkgsource::Source::Dir(dir.into()), trusted_key)?;
verify_package_source(dir, &pkgsource::Source::Dir(dir.into()), trusted_key)
}
pub fn verify_install_package(dir: &Path, trusted_key: &str) -> Result<()> {
verify_package_source(dir, &pkgsource::Source::Installation(Box::new(pkgsource::Source::Dir(dir.into()))), trusted_key)
}
fn verify_package_source(dir: &Path, source: &pkgsource::Source, trusted_key: &str) -> Result<()> {
let (manifest, _) = pkgsource::fetch_manifest(source, trusted_key)?;
if !manifest.files.iter().any(|f| f.path == schema_check::CONTRACT && f.size > 0) {
return Err(Error::Invalid("The release is missing its database schema contract.".into()));
}
@@ -249,6 +319,34 @@ mod tests {
use base64::{engine::general_purpose::STANDARD, Engine};
use ed25519_dalek::{Signer, SigningKey};
#[test]
#[ignore = "Requires a composed disposable installation image"]
fn real_full_package_installs_the_current_version_without_an_update_transaction() {
let image = std::path::PathBuf::from(std::env::var_os("COA_INSTALL_IMAGE_DIR").expect("COA_INSTALL_IMAGE_DIR"));
let provenance: serde_json::Value = fsx::read_json(&image.join("Settings/install-source.json")).unwrap();
let root = tempfile::tempdir().unwrap();
let package = root.path().join("package");
let options = package::BuildOptions {
kind: Kind::Base, version: provenance["version"].as_str().unwrap().into(),
core_commit: provenance["core"]["commit"].as_str().map(String::from),
bots_commit: provenance["bots"]["commit"].as_str().map(String::from),
built_at: "qualification".into(), part_size: package::DEFAULT_PART_SIZE, migrations: vec![],
};
package::build(&image, &package, &options, &|_| {}).unwrap();
let signing_key = SigningKey::generate(&mut rand_core::OsRng);
let bytes = fs::read(package.join("manifest.json")).unwrap();
fs::write(package.join("manifest.json.sig"), STANDARD.encode(signing_key.sign(&bytes).to_bytes())).unwrap();
let trusted_key = STANDARD.encode(signing_key.verifying_key().to_bytes());
let fixture = root.path().join("coa-schema-fixture-clean-install");
let report = qualify_install(&package, &fixture, &trusted_key).unwrap();
assert_eq!(report["version"], provenance["version"]);
assert_eq!(report["result"], "passed");
let metadata = crate::registry::metadata_dir_for(&fixture).unwrap();
let (_, installed) = crate::registry::MetaDir::open(&metadata).unwrap();
assert_eq!(installed.core.version.as_deref(), provenance["version"].as_str());
assert!(!metadata.join("updates").read_dir().is_ok_and(|mut entries| entries.next().is_some()));
}
#[test]
fn relative_startup_fixture_targets_replace_the_original_location() {
let directory = tempfile::Builder::new().prefix("coa-schema-fixture-path-test-").tempdir_in(".").unwrap();
+48 -28
View File
@@ -30,7 +30,9 @@ pub fn check(root: &Path, dir: &Path) -> Result<Vec<DatabaseCheck>> {
for mode in modes {
let db = db.clone().for_realm(mode);
let migrations: Vec<_> = m.migrations.iter().filter(|m| mode == crate::realms::Mode::Coa || m.db != "auth").cloned().collect();
checks.push(DatabaseCheck { realm: mode, migrations: crate::migrations::status(&db, &migrations)?, problems: crate::schema_check::check(&db, root)?, full_schema: root.join(crate::schema_check::CONTRACT).exists() });
let mut problems = crate::schema_check::check(&db, root)?;
problems.extend(crate::squid::database_problems(root, mode, &|sql| db.query(sql))?);
checks.push(DatabaseCheck { realm: mode, migrations: crate::migrations::status(&db, &migrations)?, problems, full_schema: root.join(crate::schema_check::CONTRACT).exists() });
}
Ok(checks)
})?;
@@ -128,38 +130,56 @@ pub fn run(root: &Path, dir: &Path, base_source: &Source, update_source: &Source
if fsx::sha256_file(&src)? != entry.sha256 { return Err(Error::Invalid("A repair file failed verification.".into())); }
}
progress("Backing up before repair", 65);
let backup = crate::backup::create(root, dir, crate::backup::Kind::Full, crate::backup::Trigger::Manual, Some("before repair".into()), &|_| {})?.id;
let backup = crate::backup::create(root, dir, crate::backup::Kind::Full, crate::backup::Trigger::BeforeRepair, Some("before repair".into()), &|phase| progress(phase, 65))?.id;
env.verify_recovery_point(&crate::backup::get(dir, &backup)?)?;
let repair_files: Vec<_> = broken.iter().map(|(file, _)| (*file).clone()).collect();
let mut txn = crate::update::begin_repair(root, dir, &update.version, &backup, &repair_files)?;
let before = dir.join("repairs").join(&backup);
fs::create_dir_all(&before)?;
fsx::require_space(root, broken.iter().map(|(e,_)| e.size.saturating_mul(2)).sum())?;
let restored = restore_files(root, &work, &before, &broken)?;
for (entry, _) in &broken {
meta.original_hashes.insert(entry.path.clone(), entry.sha256.clone());
}
fsx::atomic_write_json(&dir.join("install.json"), &meta)?;
progress("Repairing pending database updates", 80);
crate::update::repair_migrating(dir, &mut txn)?;
let (applied, error) = match env.migrate(&update, &work.join("update/_migrations")) {
Ok(r) => (r.applied, r.failed.map(|(id,why)| format!("{id}: {why}"))),
Err(e) => (Vec::new(), Some(e.to_string())),
};
if error.is_some() {
crate::update::finish_repair(root, dir, &mut txn, false, &env)?;
let report = Report { restored: Vec::new(), applied: Vec::new(), backup, database: check(root, dir)?, error };
let attempt = (|| -> Result<Report> {
let before = dir.join("repairs").join(&backup);
fs::create_dir_all(&before)?;
fsx::require_space(root, broken.iter().map(|(e,_)| e.size.saturating_mul(2)).sum())?;
let restored = restore_files(root, &work, &before, &broken)?;
for (entry, _) in &broken {
meta.original_hashes.insert(entry.path.clone(), entry.sha256.clone());
}
fsx::atomic_write_json(&dir.join("install.json"), &meta)?;
progress("Repairing pending database updates", 80);
crate::update::repair_migrating(dir, &mut txn)?;
let (applied, error) = match env.migrate(&update, &work.join("update/_migrations")) {
Ok(r) => (r.applied, r.failed.map(|(id,why)| format!("{id}: {why}"))),
Err(e) => (Vec::new(), Some(e.to_string())),
};
if error.is_some() {
crate::update::finish_repair(root, dir, &mut txn, false, &env)?;
let report = Report { restored: Vec::new(), applied: Vec::new(), backup: backup.clone(), database: check(root, dir)?, error };
fsx::atomic_write_json(&before.join("report.json"), &report)?;
return Ok(report);
}
crate::config::create_missing_module_configs(root)?;
progress("Checking database after repair", 95);
let database = check(root, dir)?;
let report = Report { restored, applied, backup: backup.clone(), database, error };
fsx::atomic_write_json(&before.join("report.json"), &report)?;
return Ok(report);
crate::update::finish_repair(root, dir, &mut txn, true, &env)?;
progress("Done", 100);
Ok(report)
})();
match attempt {
Ok(report) => Ok(report),
Err(error) if matches!(txn.state, crate::update::State::RolledBack | crate::update::State::Failed) => Err(error),
Err(error) => {
progress("Undoing the failed repair", 95);
let message = match crate::update::finish_repair(root, dir, &mut txn, false, &env) {
Ok(()) => format!("Repair failed: {error}. The original server files and databases were restored from recovery point {backup}."),
Err(recovery) => format!("Repair failed: {error}. Recovery failed: {recovery}. Recovery point {backup} was preserved; resolve the unfinished repair before starting the server."),
};
let failed = Report { restored: Vec::new(), applied: Vec::new(), backup: backup.clone(), database: Vec::new(), error: Some(message.clone()) };
if let Err(report_error) = fsx::atomic_write_json(&dir.join("repairs").join(&backup).join("report.json"), &failed) {
tracing::warn!(%report_error, "Could not persist failed repair report");
}
Err(Error::Invalid(message))
}
}
crate::config::create_missing_module_configs(root)?;
progress("Checking database after repair", 95);
let database = check(root, dir)?;
let report = Report { restored, applied, backup, database, error };
fsx::atomic_write_json(&before.join("report.json"), &report)?;
crate::update::finish_repair(root, dir, &mut txn, true, &env)?;
progress("Done", 100);
Ok(report)
})();
// Only the UUID staging directory created above is discarded; backups are retained.
let _ = fs::remove_dir_all(work);
+56
View File
@@ -1,6 +1,35 @@
use std::path::Path;
use serde::{Deserialize, Serialize};
/// SQUID has its own import ledger, outside the signed package migration ledger.
/// Keep this read-only: a pending base import can contain player data and must not
/// be replayed or cleared merely because the normal schema contract passes.
pub(crate) fn database_problems(root: &Path, mode: crate::realms::Mode, query: &impl Fn(&str) -> crate::Result<String>) -> crate::Result<Vec<crate::schema_check::Problem>> {
let active = root.join("Core/configs/modules/playerbots.conf");
let config = if active.is_file() { active } else { root.join("Core/configs/modules/playerbots.conf.dist") };
let enabled = if config.is_file() {
let conf = crate::config::parser::ConfFile::parse_bytes(&std::fs::read(config)?)?;
!matches!(conf.get("AiPlayerbot.Enabled").unwrap_or("1").trim().trim_matches('"').to_ascii_lowercase().as_str(), "0" | "false" | "no" | "off")
} else { false };
let mut problems = Vec::new();
for schema in [mode.schema("characters")?, mode.schema("world")?, "acore_playerbots"] {
let tables: std::collections::BTreeSet<_> = query(&format!("SELECT TABLE_NAME FROM information_schema.TABLES WHERE TABLE_SCHEMA='{schema}';"))?.lines().map(str::to_owned).collect();
if tables.contains("coa_squid_pending") {
for pending in query(&format!("SELECT path FROM `{schema}`.coa_squid_pending;"))?.lines().filter(|path| !path.is_empty()) {
problems.push(crate::schema_check::Problem { database: schema.into(), table: "coa_squid_pending".into(), column: String::new(), detail: format!("SQUID base import was interrupted: {pending}. File repair and package SQL updates cannot complete this import. Restore a verified database backup from before the import; do not delete the pending record or replay base SQL.") });
}
}
if enabled && schema == mode.schema("characters")? {
for table in ["playerbots_names", "playerbots_guild_names", "playerbots_arena_team_names"] {
if !tables.contains(table) {
problems.push(crate::schema_check::Problem { database: schema.into(), table: table.into(), column: String::new(), detail: "Required SQUID character table is missing. A normal server start provisions it when no interrupted import is present; an interrupted import requires database recovery first.".into() });
}
}
}
}
Ok(problems)
}
#[derive(Debug, Serialize, PartialEq, Eq)]
pub struct Release {
pub tag: Option<String>,
@@ -129,6 +158,33 @@ pub fn validate(field: &Field, value: &str) -> crate::Result<()> {
mod tests {
use super::*;
#[test]
fn database_check_reports_pending_squid_imports_in_each_realm_without_writes() {
let root = tempfile::tempdir().unwrap();
for mode in [crate::realms::Mode::Coa, crate::realms::Mode::Wildcard] {
let problems = database_problems(root.path(), mode, &|sql| {
assert!(sql.starts_with("SELECT "));
if sql.contains("information_schema.TABLES") { Ok("coa_squid_pending".into()) }
else { Ok("playerbots/base/ai_playerbot_texts.sql".into()) }
}).unwrap();
assert_eq!(problems.len(), 3);
assert_eq!(problems[0].database, mode.schema("characters").unwrap());
assert!(problems.iter().all(|p| p.detail.contains("interrupted")));
}
}
#[test]
fn enabled_squid_requires_character_tables_but_disabled_squid_does_not() {
let root = tempfile::tempdir().unwrap();
let config = root.path().join("Core/configs/modules/playerbots.conf");
crate::fsx::atomic_write(&config, b"AiPlayerbot.Enabled = 1\n").unwrap();
let query = |_: &str| Ok(String::new());
assert_eq!(database_problems(root.path(), crate::realms::Mode::Wildcard, &query).unwrap().len(), 3);
crate::fsx::atomic_write(&config, b"AiPlayerbot.Enabled = 0\n").unwrap();
assert!(database_problems(root.path(), crate::realms::Mode::Wildcard, &query).unwrap().is_empty());
crate::fsx::atomic_write(&config, b"AiPlayerbot.Enabled = 1\n").unwrap();
assert!(database_problems(root.path(), crate::realms::Mode::Coa, &|_| Ok("playerbots_names\nplayerbots_guild_names\nplayerbots_arena_team_names".into())).unwrap().is_empty());
}
#[test]
fn a_setting_with_choices_only_accepts_one_of_them() {
let root = tempfile::tempdir().unwrap();
crate::fsx::atomic_write(&root.path().join("Core/configs/modules/playerbots.conf.settings.json"), br#"{"format":1,"settings":[
+262 -184
View File
@@ -92,8 +92,6 @@ pub trait Env {
fn ensure_stopped(&self) -> Result<()>;
fn preflight(&self, _manifest: &Manifest) -> Result<()> { Ok(()) }
fn verify_snapshot(&self, _id: &str) -> Result<()> { Ok(()) }
fn rehearse(&self, _manifest: &Manifest, _tree: &Path, _items: &[PlanItem], _point: &str) -> Result<()> { Ok(()) }
/// Create a recovery point for the databases and configuration; returns its id.
fn snapshot(&self) -> Result<String>;
fn restore_snapshot(&self, id: &str) -> Result<()>;
fn migrate(&self, manifest: &Manifest, staged_migrations: &Path) -> Result<ApplyReport>;
@@ -121,13 +119,32 @@ pub fn load(meta: &Path, id: &str) -> Result<Txn> {
}
fn read_journal(path: &Path) -> Result<Txn> {
let txn: Txn = fsx::read_json(path).map_err(|e| Error::Invalid(format!("Update journal {} cannot be read: {e}. Available recovery copies were preserved.", path.display())))?;
let mut txn: Txn = fsx::read_json(path).map_err(|e| Error::Invalid(format!("Update journal {} cannot be read: {e}. Available recovery copies were preserved.", path.display())))?;
if path.parent().and_then(Path::file_name).and_then(|id| id.to_str()) != Some(txn.id.as_str()) { return Err(Error::Invalid("Update journal identity is inconsistent.".into())); }
txn_dir(path.parent().unwrap(), &txn.id)?;
if let Some(id) = &txn.recovery_point {
if id.is_empty() || !id.chars().all(|c| c.is_ascii_alphanumeric() || matches!(c, '-' | '_' | '.')) || id.contains("..") { return Err(Error::Invalid("Update journal contains an invalid recovery point identity.".into())); }
}
let valid_hash = |hash: &str| hash.len() == 64 && hash.bytes().all(|b| b.is_ascii_hexdigit());
let mut completed_legacy_repair = false;
// Older successful repairs saved Committed without setting done. Recover only that
// specific shape, and only if every repaired file still has its expected bytes.
if txn.state == State::Committed && txn.id.contains("-repair-")
&& txn.message.as_deref() == Some("Repair is in progress.")
&& txn.from_version.as_deref() == Some(txn.to_version.as_str())
&& txn.databases_started && txn.recovery_point.is_some() && txn.recovery_hashes.contains_key("@install")
&& txn.ops.iter().any(|op| !op.done)
&& txn.ops.iter().all(|op| op.started && matches!((op.action, op.had_previous), (Action::Create, false) | (Action::Replace, true)) && valid_hash(&op.new_sha256)) {
let meta_dir = path.parent().and_then(Path::parent).and_then(Path::parent)
.ok_or_else(|| Error::Invalid("Invalid repair journal location.".into()))?;
let (_, installed) = MetaDir::open(meta_dir)?;
let root = Path::new(&installed.server_path);
if fsx::canonicalize_lenient(&crate::registry::metadata_dir_for(root)?)? != fsx::canonicalize_lenient(meta_dir)? {
return Err(Error::Invalid("The repair journal does not belong to its installed server.".into()));
}
complete_repair_ops(root, &mut txn)?;
completed_legacy_repair = true;
}
let mut seen = std::collections::BTreeSet::new();
for op in &txn.ops {
fsx::safe_join(path.parent().unwrap(), &op.path)?;
@@ -140,6 +157,13 @@ fn read_journal(path: &Path) -> Result<Txn> {
}
if txn.recovery_hashes.values().any(|hash| !valid_hash(hash)) { return Err(Error::Invalid("Update journal contains invalid recovery hashes.".into())); }
for key in txn.recovery_hashes.keys().filter(|key| !matches!(key.as_str(), "@install" | "@point")) { fsx::safe_join(path.parent().unwrap(), key)?; }
if completed_legacy_repair {
// Persist the verified completion once: later legitimate updates can change
// these files. Retain the original record for diagnostics and crash recovery.
let original = path.with_file_name("txn-before-repair-completion.json");
if !original.exists() { fsx::atomic_write(&original, &fs::read(path)?)?; }
fsx::atomic_write_json(path, &txn)?;
}
Ok(txn)
}
@@ -269,6 +293,8 @@ pub fn plan(root: &Path, meta: &InstallMeta, manifest: &Manifest, resolutions: &
pub struct Preview {
pub from_version: Option<String>,
pub to_version: String,
#[serde(skip_serializing_if = "Option::is_none")]
pub latest_version: Option<String>,
pub items: Vec<PlanItem>,
pub conflicts: Vec<String>,
pub migrations: usize,
@@ -308,14 +334,17 @@ pub fn preview(root: &Path, meta: &InstallMeta, source: &Source, trusted_key: &s
pub fn preview_with(root: &Path, meta: &InstallMeta, source: &Source, trusted_key: &str, resolutions: &BTreeMap<String, Resolution>, access: DatabaseAccess) -> Result<Preview> {
refuse_for_docker(root)?;
let meta_dir = crate::registry::metadata_dir_for(root)?;
let _lock = operation_lock(&meta_dir)?;
ensure_recovered(&meta_dir)?;
let observed_version = MetaDir::open(&meta_dir)?.1.core.version;
let (m, _) = fetch_manifest(source, trusted_key)?;
let _lock = operation_lock_for(&meta_dir, "checking updates", true)?;
ensure_recovered(&meta_dir)?;
let (_, current_meta) = MetaDir::open(&meta_dir)?;
let meta = if current_meta.core.version != observed_version { &current_meta } else { meta };
check_manifest(&m)?;
reject_downgrade(meta, &m)?;
let mut items = plan(root, meta, &m, resolutions, None)?;
if items.iter().any(|item| item.path == "Scripts/manage.py" && item.action != Action::Skip)
&& launcher_already_integrated(root, &meta_dir, meta, &m, source)? {
&& launcher_already_integrated(root, meta, &m)? {
if let Some(item) = items.iter_mut().find(|item| item.path == "Scripts/manage.py") {
item.action = Action::Skip;
item.reason = Some("already up to date with Manager integration".into());
@@ -334,6 +363,7 @@ pub fn preview_with(root: &Path, meta: &InstallMeta, source: &Source, trusted_ke
Ok(Preview {
from_version: meta.core.version.clone(),
to_version: m.version.clone(),
latest_version: None,
conflicts: items.iter().filter(|i| i.action == Action::Conflict).map(|i| i.path.clone()).collect(),
items,
migrations: m.migrations.len(),
@@ -344,27 +374,11 @@ pub fn preview_with(root: &Path, meta: &InstallMeta, source: &Source, trusted_ke
})
}
fn launcher_already_integrated(root: &Path, meta_dir: &Path, meta: &InstallMeta, manifest: &Manifest, source: &Source) -> Result<bool> {
fn launcher_already_integrated(root: &Path, meta: &InstallMeta, manifest: &Manifest) -> Result<bool> {
let Some(file) = manifest.files.iter().find(|file| file.path == "Scripts/manage.py") else { return Ok(false); };
let actual = sha_if_exists(&root.join(&file.path));
if actual.as_ref() != meta.original_hashes.get(&file.path) || actual.is_none() { return Ok(false); }
let cache = meta_dir.join("manifests").join(format!("launcher-{}.py", file.sha256));
let signed = match fs::read(&cache) {
Ok(bytes) if fsx::sha256_bytes(&bytes) == file.sha256 => bytes,
_ => {
let temp = tempfile::tempdir()?;
let parts = fetch_parts(source, manifest, &temp.path().join("download"), &Cancel::default(), &|_, _| {})?;
package::extract(&parts, manifest, &temp.path().join("tree"), &|_, _| {})?;
let bytes = fs::read(temp.path().join("tree").join(&file.path))?;
if fsx::sha256_bytes(&bytes) != file.sha256 { return Err(Error::Invalid("The signed launcher failed verification.".into())); }
// This derived cache is optional; failure to save it does not change the server.
let _ = fsx::atomic_write(&cache, &bytes);
bytes
}
};
Ok(std::str::from_utf8(&signed).ok().is_some_and(|signed| {
fs::read(root.join(&file.path)).ok().is_some_and(|actual| crate::driver::launcher_matches(signed, &actual))
}))
let Ok(actual) = fs::read(root.join(&file.path)) else { return Ok(false); };
if meta.original_hashes.get(&file.path) != Some(&fsx::sha256_bytes(&actual)) { return Ok(false); }
Ok(crate::driver::launcher_matches_recorded(&file.sha256, &actual))
}
fn reject_downgrade(meta: &InstallMeta, manifest: &Manifest) -> Result<()> {
@@ -373,8 +387,10 @@ fn reject_downgrade(meta: &InstallMeta, manifest: &Manifest) -> Result<()> {
if c.platform != platform {
return Err(Error::Invalid(format!("This package targets {} rather than {platform}.", c.platform)));
}
if meta.core.version.as_ref().is_none_or(|version| !c.source_versions.contains(version)) {
return Err(Error::Invalid("This server version is outside the release's tested compatibility matrix.".into()));
// The source list records release qualification, not an exclusive upgrade route.
// Database history and target validation decide whether the update can complete.
if meta.core.version.as_deref().and_then(crate::manifest::parse_version).is_none() {
return Err(Error::Invalid("The installed server version is unknown. An imported repack needs a verified starting state before updating.".into()));
}
}
if let Some(installed) = meta.core.version.as_deref() {
@@ -495,14 +511,6 @@ pub fn apply(p: &Params, report: &dyn Fn(&str, u8)) -> Result<Outcome> {
let point_meta = crate::backup::point_json(meta_dir, txn.recovery_point.as_deref().unwrap())?;
if point_meta.is_file() { txn.recovery_hashes.insert("@point".into(), fsx::sha256_file(&point_meta)?); }
save(meta_dir, &txn)?;
step(report, "Testing the update on a private copy of your server", 54);
if let Err(error) = p.env.rehearse(&m, &tree, &items, txn.recovery_point.as_deref().unwrap()) {
txn.state = State::Failed;
txn.message = Some(format!("The private update rehearsal failed: {error}. The installed server was not updated."));
save(meta_dir, &txn)?;
return Err(Error::Invalid(txn.message.unwrap()));
}
txn.ops = items
.iter()
.filter(|i| i.action != Action::Skip)
@@ -780,11 +788,26 @@ pub fn rollback(root: &Path, meta_dir: &Path, id: &str, env: &dyn Env) -> Result
}
pub(crate) fn operation_lock(meta_dir: &Path) -> Result<fs::File> {
operation_lock_for(meta_dir, "server maintenance", false)
}
pub(crate) fn operation_lock_for(meta_dir: &Path, operation: &str, wait_for_check: bool) -> Result<fs::File> {
fs::create_dir_all(meta_dir)?;
let lock = fs::OpenOptions::new().read(true).write(true).create(true).truncate(false).open(meta_dir.join("update.lock"))?;
if !fs4::fs_std::FileExt::try_lock_exclusive(&lock)? {
return Err(Error::Invalid("Another update, repair or recovery is in progress.".into()));
let deadline = std::time::Instant::now() + std::time::Duration::from_secs(30);
loop {
if fs4::fs_std::FileExt::try_lock_exclusive(&lock)? { break; }
let owner = fsx::read_json::<serde_json::Value>(&meta_dir.join("update-lock-owner.json")).ok();
let checking = owner.as_ref().and_then(|owner| owner.get("operation")).and_then(|value| value.as_str()) == Some("checking updates");
if wait_for_check && checking && std::time::Instant::now() < deadline {
std::thread::sleep(std::time::Duration::from_millis(100));
continue;
}
let details = owner.as_ref().and_then(|owner| Some(format!(" Operation: {}; Manager process: {}; started: {}.", owner.get("operation")?.as_str()?, owner.get("pid")?.as_u64()?, owner.get("started")?.as_str()?))).unwrap_or_default();
return Err(Error::Invalid(format!("Another update, repair or recovery is in progress.{details} Wait for that operation to finish, then retry.")));
}
let record = serde_json::json!({"operation": operation, "pid": std::process::id(), "started": chrono::Utc::now().to_rfc3339()});
fsx::atomic_write_json(&meta_dir.join("update-lock-owner.json"), &record)?;
Ok(lock)
}
@@ -802,6 +825,8 @@ pub(crate) fn begin_repair(root: &Path, meta_dir: &Path, version: &str, backup:
}
let mut recovery_hashes = BTreeMap::new();
recovery_hashes.insert("@install".into(), fsx::sha256_file(&before.join("manager-install.json"))?);
let point = crate::backup::point_json(meta_dir, backup)?;
if point.is_file() { recovery_hashes.insert("@point".into(), fsx::sha256_file(&point)?); }
for op in ops.iter().filter(|op| op.had_previous) { recovery_hashes.insert(op.path.clone(), fsx::sha256_file(&fsx::safe_join(&before, &op.path)?)?); }
let txn = Txn { id, state: State::Applying, from_version: Some(version.into()), to_version: version.into(), recovery_point: Some(backup.into()), databases_started: false, ops, message: Some("Repair is in progress.".into()), recovery_hashes };
save(meta_dir, &txn)?;
@@ -815,11 +840,30 @@ pub(crate) fn repair_migrating(meta_dir: &Path, txn: &mut Txn) -> Result<()> {
}
pub(crate) fn finish_repair(root: &Path, meta_dir: &Path, txn: &mut Txn, success: bool, env: &dyn Env) -> Result<()> {
if !success { restore_transaction(root, meta_dir, &txn_dir(meta_dir, &txn.id)?.join("before"), txn, env)?; }
if !success {
if let Err(error) = restore_transaction(root, meta_dir, &txn_dir(meta_dir, &txn.id)?.join("before"), txn, env) {
txn.state = State::Failed;
txn.message = Some(format!("Repair recovery failed: {error}. Its recovery point was preserved."));
save(meta_dir, txn)?;
return Err(error);
}
}
if success { complete_repair_ops(root, txn)?; }
txn.state = if success { State::Committed } else { State::RolledBack };
save(meta_dir, txn)
}
fn complete_repair_ops(root: &Path, txn: &mut Txn) -> Result<()> {
for op in &txn.ops {
let target = fsx::ensure_within(root, &fsx::safe_join(root, &op.path)?)?;
if fsx::sha256_file(&target)? != op.new_sha256 {
return Err(Error::Invalid(format!("Repaired file {} failed verification. Its recovery record was preserved.", op.path)));
}
}
for op in &mut txn.ops { op.done = true; }
Ok(())
}
fn prune(meta_dir: &Path, keep: usize) {
let Ok(rd) = fs::read_dir(updates_dir(meta_dir)) else { return };
let mut dirs: Vec<_> = rd.flatten().filter(|e| read_journal(&e.path().join("txn.json")).is_ok_and(|t| matches!(t.state, State::Committed | State::RolledBack))).map(|e| e.path()).collect();
@@ -854,7 +898,7 @@ impl RepackEnv<'_> {
if realms.wildcard_created { modes.push(if realms.active == crate::realms::Mode::Coa { crate::realms::Mode::Wildcard } else { crate::realms::Mode::Coa }); }
let mut count = 0;
for (index, mode) in modes.into_iter().enumerate() {
let list: Vec<_> = manifest.migrations.iter().filter(|m| index == 0 || m.db != "auth").cloned().collect();
let list: Vec<_> = manifest.migrations.iter().filter(|m| index == 0 || !crate::migrations::shared_auth(&m.db)).cloned().collect();
count += crate::migrations::pending_count(&db.clone().for_realm(mode), &list)?;
}
Ok(count)
@@ -920,77 +964,6 @@ impl Env for RepackEnv<'_> {
}
fn automatic_rollback(&self) -> bool { true }
fn rehearse(&self, manifest: &Manifest, tree: &Path, items: &[PlanItem], id: &str) -> Result<()> {
if items.iter().any(|item| item.action != Action::Skip && item.path.to_ascii_lowercase().starts_with("mysql/")) {
return Err(Error::Invalid("A server update cannot replace MySQL while using a cold recovery copy; a separate database engine upgrade is required.".into()));
}
let point = crate::backup::get(self.meta_dir, id)?;
let Some(snapshot) = &point.mysql_snapshot else {
// Docker uses its separate experimental release channel until its clone runner is qualified.
if crate::docker::is_docker(self.root) { return Ok(()); }
return Err(Error::Invalid("This installation has no complete MySQL recovery copy for an update rehearsal.".into()));
};
crate::mysql_snapshot::validate_connections(self.root, self.root)?;
crate::mysql_snapshot::stop(self.root)?;
let point_dir = crate::backup::point_json(self.meta_dir, id)?.parent().unwrap().to_path_buf();
let fixture = crate::mysql_snapshot::rehearsal_copy(self.root, self.meta_dir, &point_dir, snapshot)?;
let metadata = crate::registry::metadata_dir_for(&fixture)?;
let env = RepackEnv { root: &fixture, meta_dir: &metadata };
let result = crate::install::with_scratch_ports(&fixture, || {
let mut txn = Txn { id: "rehearsal".into(), state: State::Applying, from_version: None,
to_version: manifest.version.clone(), recovery_point: None, databases_started: false,
ops: items.iter().filter(|item| item.action != Action::Skip).map(|item| Op {
path: item.path.clone(), action: item.action, reason: item.reason.clone(),
new_sha256: manifest.files.iter().find(|file| file.path == item.path).unwrap().sha256.clone(),
had_previous: fsx::safe_join(&fixture, &item.path).is_ok_and(|path| path.is_file()), started:false, done:false,
}).collect(), message:None, recovery_hashes:BTreeMap::new() };
let before = txn_dir(&metadata, &txn.id)?.join("before");
fs::create_dir_all(&before)?;
apply_ops(&fixture, tree, &before, &manifest.files, &mut txn, &metadata, None)?;
crate::mysql_snapshot::validate_connections(&fixture, self.root)?;
crate::mysql_snapshot::isolate_connections(&fixture)?;
let identities = crate::backup::with_database(&fixture, |db| {
let identities = database_identities(db)?;
let realms = crate::realms::state(&fixture)?;
let mut modes = vec![realms.active];
if realms.wildcard_created { modes.push(if realms.active == crate::realms::Mode::Coa { crate::realms::Mode::Wildcard } else { crate::realms::Mode::Coa }); }
for (index, mode) in modes.into_iter().enumerate() {
let db = db.clone().for_realm(mode);
crate::schema_check::repair_missing_defaults(&db, &fixture, manifest.files.iter().find(|file| file.path == crate::schema_check::CONTRACT).map(|file| file.sha256.as_str()))?;
let list: Vec<_> = manifest.migrations.iter().filter(|migration| index == 0 || migration.db != "auth").cloned().collect();
let report = crate::migrations::apply_pending(&db, &list, &tree.join("_migrations"), &|| Ok("private rehearsal".into()))?;
if let Some((migration, error)) = report.failed { return Err(Error::Invalid(format!("Rehearsal migration {migration} failed: {error}"))); }
crate::schema_check::repair_missing_defaults(&db, &fixture, manifest.files.iter().find(|file| file.path == crate::schema_check::CONTRACT).map(|file| file.sha256.as_str()))?;
}
verify_database_identities(db, &identities)?;
Ok(identities)
})?;
env.validate()?;
crate::backup::with_database(&fixture, |db| verify_database_identities(db, &identities))?;
env.ensure_stopped()
});
let stopped = crate::mysql_snapshot::stop(&fixture);
let live_state = if result.is_ok() { crate::mysql_snapshot::stop(self.root).and_then(|()| crate::mysql_snapshot::unchanged(self.root, snapshot)) } else { Ok(()) };
let logs: BTreeMap<_, _> = ["Core/Logs/auth-console.log", "Core/Logs/world-console.log", "Core/Logs/supervisor.log", "mysql/logs/mysql-error.log"].into_iter()
.map(|path| (path, crate::diag::redact(&crate::health::tail(&fixture.join(path), 32768)))).collect();
let mut schema_reports = Vec::<serde_json::Value>::new();
if let Ok(entries) = fs::read_dir(metadata.join("diagnostics")) {
for entry in entries.flatten().filter(|entry| entry.file_name().to_string_lossy().starts_with("schema-validation-")) {
let path = fsx::ensure_within(&metadata, &entry.path())?;
if fs::metadata(&path)?.len() <= 4 * 1024 * 1024 { schema_reports.push(fsx::read_json(&path)?); }
}
}
let report = serde_json::json!({"schema":1,"checkedAt":chrono::Utc::now().to_rfc3339(),"toVersion":manifest.version,"recoveryPoint":id,
"result":if result.is_ok() && stopped.is_ok() && live_state.is_ok() {"passed"} else {"failed"}, "error":result.as_ref().err().map(ToString::to_string), "stopError":stopped.as_ref().err().map(ToString::to_string), "liveStateError":live_state.as_ref().err().map(ToString::to_string), "logs":logs,"schemaReports":schema_reports,"fixture":fixture});
let report_path = self.meta_dir.join("diagnostics").join(format!("database-rehearsal-{}.json", uuid::Uuid::new_v4()));
fsx::atomic_write_json(&report_path, &report)?;
stopped?;
let parent = fixture.parent().unwrap();
let owned = fsx::ensure_within(&self.meta_dir.join("rehearsals"), parent)?;
fs::remove_dir_all(owned)?;
live_state?;
result
}
fn preflight(&self, manifest: &Manifest) -> Result<()> {
crate::mysql_snapshot::stop_previous_rehearsals(self.meta_dir)?;
if let Some(compatibility) = &manifest.compatibility {
@@ -1010,7 +983,7 @@ impl Env for RepackEnv<'_> {
if crate::docker::is_docker(self.root) {
return Err(Error::Invalid(format!("The installed {} database does not match its upgrade source: {}.{}: {}", mode.name(), problem.database, problem.table, problem.detail)));
}
tracing::warn!(realm = mode.name(), table = %problem.table, "Installed schema differs; the private candidate rehearsal must qualify this database");
tracing::warn!(realm = mode.name(), table = %problem.table, "Installed schema differs; pending migrations and target validation will run with a full recovery point");
}
}
Ok(())
@@ -1065,43 +1038,10 @@ impl Env for RepackEnv<'_> {
}
fn migrate(&self, manifest: &Manifest, staged: &Path) -> Result<ApplyReport> {
let root = self.root;
crate::backup::with_database(root, |db| {
let realms = crate::realms::state(root)?;
if let Some(contract) = manifest.files.iter().find(|file| file.path == crate::schema_check::CONTRACT) {
self.migration_snapshot()?;
crate::schema_check::repair_missing_defaults(db, root, Some(&contract.sha256))?;
}
let mut result = crate::migrations::apply_pending(db, &manifest.migrations, staged, &|| {
self.migration_snapshot()
})?;
if realms.wildcard_created && result.failed.is_none() {
let other = if realms.active == crate::realms::Mode::Coa { crate::realms::Mode::Wildcard } else { crate::realms::Mode::Coa };
let other_db = db.clone().for_realm(other);
crate::schema_check::repair_missing_defaults(&other_db, root, manifest.files.iter().find(|file| file.path == crate::schema_check::CONTRACT).map(|file| file.sha256.as_str()))?;
let shared_snapshot = result.snapshot.clone();
let migrations: Vec<_> = manifest.migrations.iter().filter(|m| m.db != "auth").cloned().collect();
let extra = crate::migrations::apply_pending(&other_db, &migrations, staged, &|| {
if let Some(id) = &shared_snapshot { return Ok(id.clone()); }
self.migration_snapshot()
})?;
result.applied.extend(extra.applied);
result.failed = extra.failed;
if result.snapshot.is_none() { result.snapshot = extra.snapshot; }
if result.failed.is_none() {
crate::schema_check::repair_missing_defaults(&other_db, root, manifest.files.iter().find(|file| file.path == crate::schema_check::CONTRACT).map(|file| file.sha256.as_str()))?;
if let Some(p) = crate::schema_check::check_with_report(&other_db, root)?.first() {
return Err(Error::Invalid(format!("Database validation failed on {}: {}.{}: {}", other.name(), p.table, p.column, p.detail)));
}
}
}
if result.failed.is_none() {
crate::schema_check::repair_missing_defaults(db, root, manifest.files.iter().find(|file| file.path == crate::schema_check::CONTRACT).map(|file| file.sha256.as_str()))?;
let problems = crate::schema_check::check_with_report(db, root)?;
if let Some(p) = problems.first() {
return Err(Error::Invalid(format!("Database validation failed: {}.{}.{}: {} ({} problems).", p.database, p.table, p.column, p.detail, problems.len())));
}
}
crate::backup::with_database(self.root, |db| {
let identities = database_identities(db)?;
let result = migrate_databases(db, self.root, manifest, staged, &|| self.migration_snapshot())?;
if result.failed.is_none() { verify_database_identities(db, &identities)?; }
Ok(result)
})
}
@@ -1117,23 +1057,24 @@ impl Env for RepackEnv<'_> {
impl RepackEnv<'_> {
fn validate_server(&self, check_schema: bool) -> Result<()> {
let stopped = crate::driver::run(self.root, crate::driver::Verb::StopAll)?;
if !stopped.ok { return Err(Error::Invalid("Could not stop the server before isolated validation.".into())); }
crate::mysql_snapshot::stop(self.root)?;
self.ensure_stopped()?;
self.recover_validation()?;
let result = (|| -> Result<()> {
crate::update_isolation::begin(self.root, self.meta_dir)?;
use crate::process::{observe, ServiceState};
if check_schema { crate::backup::with_database(self.root, |db| {
let mut modes = vec![crate::realms::Mode::Coa];
if crate::realms::state(self.root)?.wildcard_created { modes.push(crate::realms::Mode::Wildcard); }
for mode in modes {
if let Some(p) = crate::schema_check::check_with_report(&db.clone().for_realm(mode), self.root)?.first() {
return Err(Error::Invalid(format!("Database validation failed on {}: {}.{}: {}", mode.name(), p.database, p.table, p.detail)));
let identities = crate::backup::with_database(self.root, |db| {
if check_schema {
let mut modes = vec![crate::realms::Mode::Coa];
if crate::realms::state(self.root)?.wildcard_created { modes.push(crate::realms::Mode::Wildcard); }
for mode in modes {
if let Some(p) = crate::schema_check::check_with_report(&db.clone().for_realm(mode), self.root)?.first() {
return Err(Error::Invalid(format!("Database validation failed on {}: {}.{}: {}", mode.name(), p.database, p.table, p.detail)));
}
}
}
Ok(())
})?; }
database_identities(db)
})?;
// The check start is the first start of the new build: give the module configs the settings the update added
// (as a normal start does), or the server logs a "missing property" line for every one of them.
match crate::registry::MetaDir::open(self.meta_dir) {
@@ -1171,6 +1112,7 @@ impl RepackEnv<'_> {
}
};
if healthy {
crate::backup::with_database(self.root, |db| verify_database_identities(db, &identities))?;
return Ok(());
}
let cause = crate::driver::startup_failure(self.root, &started);
@@ -1178,10 +1120,7 @@ impl RepackEnv<'_> {
let _ = crate::driver::run(self.root, crate::driver::Verb::StopAll);
Err(Error::Invalid(cause))
})();
let stopped = crate::driver::run(self.root, crate::driver::Verb::StopAll)?;
if !stopped.ok {
return Err(Error::Invalid("Isolated validation could not stop the server; recovery configurations were preserved.".into()));
}
crate::mysql_snapshot::stop(self.root)?;
self.ensure_stopped()?;
self.recover_validation()?;
result
@@ -1189,6 +1128,40 @@ impl RepackEnv<'_> {
}
/// One migration pass for both realms, with shared history checks and one recovery point.
fn migrate_databases(db: &crate::db::Db, root: &Path, manifest: &Manifest, staged: &Path, snapshot: &dyn Fn() -> Result<String>) -> Result<ApplyReport> {
let realms = crate::realms::state(root)?;
let mut modes = vec![realms.active];
if realms.wildcard_created {
modes.push(if realms.active == crate::realms::Mode::Coa { crate::realms::Mode::Wildcard } else { crate::realms::Mode::Coa });
}
// Reject a broken history in either realm before changing the first one.
for (index, mode) in modes.iter().enumerate() {
let list: Vec<_> = manifest.migrations.iter().filter(|m| index == 0 || !crate::migrations::shared_auth(&m.db)).cloned().collect();
crate::migrations::preflight(&db.clone().for_realm(*mode), &list)?;
}
let contract = manifest.files.iter().find(|file| file.path == crate::schema_check::CONTRACT).map(|file| file.sha256.as_str());
let mut result = ApplyReport { applied: vec![], failed: None, snapshot: None };
if contract.is_some() { result.snapshot = Some(snapshot()?); }
for (index, mode) in modes.into_iter().enumerate() {
let db = db.clone().for_realm(mode);
let list: Vec<_> = manifest.migrations.iter().filter(|m| index == 0 || !crate::migrations::shared_auth(&m.db)).cloned().collect();
crate::schema_check::repair_missing_defaults(&db, root, contract)?;
let extra = crate::migrations::apply_pending(&db, &list, staged, &|| {
if let Some(id) = &result.snapshot { Ok(id.clone()) } else { snapshot() }
})?;
result.applied.extend(extra.applied);
result.failed = extra.failed;
if result.snapshot.is_none() { result.snapshot = extra.snapshot; }
if result.failed.is_some() { return Ok(result); }
crate::schema_check::repair_missing_defaults(&db, root, contract)?;
if let Some(p) = crate::schema_check::check_with_report(&db, root)?.first() {
return Err(Error::Invalid(format!("Database validation failed on {}: {}.{}: {}", mode.name(), p.table, p.column, p.detail)));
}
}
Ok(result)
}
fn database_identities(db: &crate::db::Db) -> Result<BTreeMap<String, std::collections::BTreeSet<String>>> {
let db = db.clone().for_realm(crate::realms::Mode::Coa);
let mut identities = BTreeMap::new();
@@ -1208,7 +1181,7 @@ fn verify_database_identities(db: &crate::db::Db, before: &BTreeMap<String, std:
let after = database_identities(db)?;
for (schema, expected) in before {
if after.get(schema).is_none_or(|actual| !expected.is_subset(actual)) {
return Err(Error::Invalid(format!("The private update changed or removed existing character or account identities in {schema}.")));
return Err(Error::Invalid(format!("The update changed or removed existing character or account identities in {schema}.")));
}
}
Ok(())
@@ -1236,19 +1209,15 @@ mod tests {
migrate_fail: Cell<bool>,
preflight_fail: Cell<bool>,
restore_fail: Cell<bool>,
rehearsal_fail: Cell<bool>,
}
impl Fake {
fn ok() -> Fake {
Fake { activation_expected: RefCell::new(None), activation_failed: Cell::new(false), activated: Cell::new(false), automatic_rollback: Cell::new(false), recovery_healthy: Cell::new(true), stopped: Cell::new(true), healthy: Cell::new(true), snapshot_ok: Cell::new(true), calls: Default::default(), migrate_fail: Cell::new(false), preflight_fail: Cell::new(false), restore_fail: Cell::new(false), rehearsal_fail: Cell::new(false) }
Fake { activation_expected: RefCell::new(None), activation_failed: Cell::new(false), activated: Cell::new(false), automatic_rollback: Cell::new(false), recovery_healthy: Cell::new(true), stopped: Cell::new(true), healthy: Cell::new(true), snapshot_ok: Cell::new(true), calls: Default::default(), migrate_fail: Cell::new(false), preflight_fail: Cell::new(false), restore_fail: Cell::new(false) }
}
}
impl Env for Fake {
fn rehearse(&self, _: &Manifest, _: &Path, _: &[PlanItem], _: &str) -> Result<()> {
if self.rehearsal_fail.get() { Err(Error::Invalid("SQL failed in the private copy".into())) } else { Ok(()) }
}
fn activate(&self) -> Result<()> {
if let Some((meta, version)) = self.activation_expected.borrow().as_ref() {
ensure_recovered(meta)?;
@@ -1358,6 +1327,25 @@ mod tests {
w
}
#[test]
fn startup_waits_for_an_update_check_but_not_for_active_maintenance() {
let temp = tempfile::tempdir().unwrap();
let checker = operation_lock_for(temp.path(), "checking updates", false).unwrap();
let root = temp.path().to_path_buf();
let release = std::thread::spawn(move || {
std::thread::sleep(std::time::Duration::from_millis(150));
drop(checker);
});
let start = operation_lock_for(&root, "starting server", true).unwrap();
release.join().unwrap();
let error = operation_lock(&root).unwrap_err().to_string();
assert!(error.contains("starting server"));
assert!(error.contains(&std::process::id().to_string()));
drop(start);
// Stale owner metadata is harmless once the OS lock is released.
operation_lock(&root).unwrap();
}
#[test]
fn legacy_manager_realm_launcher_is_replaced_during_update() {
let w = legacy_launcher_world(false);
@@ -1488,6 +1476,8 @@ mod tests {
let (_, mut meta) = MetaDir::open(&w.meta).unwrap();
meta.original_hashes.insert("Scripts/manage.py".into(), fsx::sha256_bytes(integrated.as_bytes()));
fsx::atomic_write_json(&w.meta.join("install.json"), &meta).unwrap();
let original = fetch_manifest(&Source::Dir(w.pkg.clone()), &w.key).unwrap().0;
for part in &original.archive.as_ref().unwrap().parts { fs::remove_file(w.pkg.join(&part.name)).unwrap(); }
let p = preview(&w.root, &meta, &Source::Dir(w.pkg.clone()), &w.key, &BTreeMap::new()).unwrap();
assert_eq!(p.items.iter().find(|item| item.path == "Scripts/manage.py").unwrap().action, Action::Skip);
let manifest = fetch_manifest(&Source::Dir(w.pkg.clone()), &w.key).unwrap().0;
@@ -1495,7 +1485,7 @@ mod tests {
assert_eq!(applied_plan.iter().find(|item| item.path == "Scripts/manage.py").unwrap().action, Action::Skip);
let mut changed = fetch_manifest(&Source::Dir(w.pkg.clone()), &w.key).unwrap().0;
changed.files.iter_mut().find(|file| file.path == "Scripts/manage.py").unwrap().sha256 = fsx::sha256_bytes(b"new launcher");
assert!(launcher_already_integrated(&w.root, &w.meta, &meta, &changed, &Source::Dir(w.pkg.clone())).is_err());
assert!(!launcher_already_integrated(&w.root, &meta, &changed).unwrap());
write(&w.root, "Scripts/manage.py", format!("{integrated}\n# user edit\n").as_bytes());
// Same version: the owner's edit is not an update (Repair lists and restores it).
let p = preview(&w.root, &meta, &Source::Dir(w.pkg.clone()), &w.key, &BTreeMap::new()).unwrap();
@@ -1642,7 +1632,7 @@ mod tests {
}
#[test]
fn incompatible_source_or_platform_is_rejected_without_touching_installation() {
fn unlisted_source_is_allowed_but_unknown_versions_and_other_platforms_are_rejected() {
let w = world(&[], false);
let (_, meta) = MetaDir::open(&w.meta).unwrap();
let mut manifest = Manifest::parse(&fs::read(w.pkg.join("manifest.json")).unwrap()).unwrap();
@@ -1655,10 +1645,19 @@ mod tests {
manifest.validate().unwrap();
validate_candidate(&meta, &manifest).unwrap();
manifest.compatibility.as_mut().unwrap().source_versions = vec!["0.9.0".into()];
assert!(validate_candidate(&meta, &manifest).unwrap_err().to_string().contains("compatibility matrix"));
validate_candidate(&meta, &manifest).unwrap();
let mut unknown = meta.clone();
unknown.core.version = None;
assert!(validate_candidate(&unknown, &manifest).unwrap_err().to_string().contains("unknown"));
unknown.core.version = Some("unverified-repack".into());
assert!(validate_candidate(&unknown, &manifest).is_err());
let mut already_updated = meta.clone();
already_updated.core.version = Some(manifest.version.clone());
validate_candidate(&already_updated, &manifest).unwrap();
manifest.compatibility.as_mut().unwrap().source_versions = vec!["1.0.0".into()];
manifest.compatibility.as_mut().unwrap().platform = if cfg!(windows) { "linux-x86_64" } else { "windows-x86_64" }.into();
assert!(validate_candidate(&meta, &manifest).unwrap_err().to_string().contains("targets"));
assert!(validate_candidate(&already_updated, &manifest).unwrap_err().to_string().contains("targets"));
assert_eq!(read(&w, "Core/worldserver.exe"), b"world-v1");
assert!(unfinished(&w.meta).is_none());
}
@@ -1835,6 +1834,88 @@ mod tests {
assert_eq!(read(&w, "Core/worldserver.exe"), b"world-v1");
}
#[test]
fn legacy_completed_repair_is_accepted_only_with_verified_repaired_files() {
let w = world(&[], false);
let files = vec![FileEntry { path: "Core/worldserver.exe".into(), sha256: fsx::sha256_bytes(b"repaired"), size: 8, owner: crate::manifest::Owner::Core, policy: ReplacePolicy::Replace }];
let mut txn = begin_repair(&w.root, &w.meta, "1.0.0", "full-backup", &files).unwrap();
write(&w.root, "Core/worldserver.exe", b"repaired");
repair_migrating(&w.meta, &mut txn).unwrap();
txn.state = State::Committed;
save(&w.meta, &txn).unwrap();
let original = fs::read(txn_dir(&w.meta, &txn.id).unwrap().join("txn.json")).unwrap();
write(&w.root, "Core/worldserver.exe", b"changed");
assert!(ensure_recovered(&w.meta).is_err());
write(&w.root, "Core/worldserver.exe", b"repaired");
ensure_recovered(&w.meta).unwrap();
assert!(load(&w.meta, &txn.id).unwrap().ops[0].done);
assert_eq!(fs::read(txn_dir(&w.meta, &txn.id).unwrap().join("txn-before-repair-completion.json")).unwrap(), original);
write(&w.root, "Core/worldserver.exe", b"a later legitimate update");
ensure_recovered(&w.meta).unwrap();
txn.ops[0].started = false;
save(&w.meta, &txn).unwrap();
assert!(ensure_recovered(&w.meta).is_err());
}
#[test]
fn changed_repair_backup_inventory_blocks_recovery_before_stopping_or_restoring() {
let w = world(&[], false);
let point = crate::backup::point_json(&w.meta, "full-backup").unwrap();
fsx::atomic_write(&point, b"original recovery inventory").unwrap();
let files = vec![FileEntry { path: "Core/worldserver.exe".into(), sha256: fsx::sha256_bytes(b"repaired"), size: 8, owner: crate::manifest::Owner::Core, policy: ReplacePolicy::Replace }];
let mut txn = begin_repair(&w.root, &w.meta, "1.0.0", "full-backup", &files).unwrap();
write(&w.root, "Core/worldserver.exe", b"repaired");
repair_migrating(&w.meta, &mut txn).unwrap();
fsx::atomic_write(&point, b"changed recovery inventory").unwrap();
let env = Fake::ok();
assert!(finish_repair(&w.root, &w.meta, &mut txn, false, &env).is_err());
assert!(env.calls.borrow().is_empty());
assert_eq!(read(&w, "Core/worldserver.exe"), b"repaired");
assert!(ensure_recovered(&w.meta).is_err());
}
#[test]
fn failed_repair_recovery_keeps_a_failed_journal_and_its_reason() {
let w = world(&[], false);
let files = vec![FileEntry { path: "Core/worldserver.exe".into(), sha256: fsx::sha256_bytes(b"repaired"), size: 8, owner: crate::manifest::Owner::Core, policy: ReplacePolicy::Replace }];
let mut txn = begin_repair(&w.root, &w.meta, "1.0.0", "full-backup", &files).unwrap();
write(&w.root, "Core/worldserver.exe", b"repaired");
repair_migrating(&w.meta, &mut txn).unwrap();
let env = Fake::ok();
env.restore_fail.set(true);
assert!(finish_repair(&w.root, &w.meta, &mut txn, false, &env).is_err());
let saved = load(&w.meta, &txn.id).unwrap();
assert_eq!(saved.state, State::Failed);
assert!(saved.message.unwrap().contains("Repair recovery failed"));
assert_eq!(saved.recovery_point.as_deref(), Some("full-backup"));
assert!(ensure_recovered(&w.meta).is_err());
}
#[test]
fn repair_cannot_commit_unverified_files() {
let w = world(&[], false);
let files = vec![FileEntry { path: "Core/worldserver.exe".into(), sha256: fsx::sha256_bytes(b"repaired"), size: 8, owner: crate::manifest::Owner::Core, policy: ReplacePolicy::Replace }];
let mut txn = begin_repair(&w.root, &w.meta, "1.0.0", "full-backup", &files).unwrap();
assert!(finish_repair(&w.root, &w.meta, &mut txn, true, &Fake::ok()).is_err());
assert_eq!(load(&w.meta, &txn.id).unwrap().state, State::Applying);
assert_eq!(pending_checked(&w.meta).unwrap().unwrap().id, txn.id);
}
#[test]
fn successful_repair_leaves_a_readable_journal_and_allows_next_operation() {
let w = world(&[], false);
let files = vec![FileEntry { path: "Core/worldserver.exe".into(), sha256: fsx::sha256_bytes(b"repaired"), size: 8, owner: crate::manifest::Owner::Core, policy: ReplacePolicy::Replace }];
let mut txn = begin_repair(&w.root, &w.meta, "1.0.0", "full-backup", &files).unwrap();
write(&w.root, "Core/worldserver.exe", b"repaired");
repair_migrating(&w.meta, &mut txn).unwrap();
finish_repair(&w.root, &w.meta, &mut txn, true, &Fake::ok()).unwrap();
let saved = load(&w.meta, &txn.id).unwrap();
assert_eq!(saved.state, State::Committed);
assert!(saved.ops.iter().all(|op| op.started && op.done));
ensure_recovered(&w.meta).unwrap();
assert_eq!(read(&w, "Core/worldserver.exe"), b"repaired");
}
#[test]
fn interrupted_repair_restores_files_databases_and_ownership_metadata() {
let w = world(&[], false);
@@ -1890,20 +1971,17 @@ mod tests {
}
#[test]
fn failed_rehearsal_preserves_installed_files_version_and_recovery_point() {
fn update_saves_recovery_then_migrates_and_validates_once() {
let w = world(&[], true);
let env = Fake::ok();
env.rehearsal_fail.set(true);
let original = fsx::sha256_file(&w.meta.join("install.json")).unwrap();
let error = run(&w, &env, BTreeMap::new(), None).unwrap_err();
assert!(error.to_string().contains("installed server was not updated"));
assert_eq!(read(&w, "Core/worldserver.exe"), b"world-v1");
assert_eq!(fsx::sha256_file(&w.meta.join("install.json")).unwrap(), original);
assert_eq!(&*env.calls.borrow(), &["stop", "snapshot"]);
assert!(pending_checked(&w.meta).unwrap().is_none());
let journal = journals(&w.meta).unwrap().remove(0);
assert_eq!(journal.recovery_point.as_deref(), Some("rp-1"));
assert!(!journal.databases_started && journal.ops.is_empty());
let result = run(&w, &env, BTreeMap::new(), None).unwrap();
assert_eq!(result.txn.state, State::Committed);
let calls = env.calls.borrow();
assert_eq!(calls.iter().filter(|call| **call == "snapshot").count(), 1);
assert_eq!(calls.iter().filter(|call| **call == "migrate").count(), 1);
assert_eq!(calls.iter().filter(|call| **call == "validate").count(), 1);
assert!(calls.iter().position(|call| *call == "snapshot") < calls.iter().position(|call| *call == "migrate"));
assert!(!w.meta.join("rehearsals").exists());
}
#[test]
+93
View File
@@ -0,0 +1,93 @@
use crate::{manifest::Manifest, pkgsource::{self, Source}, registry::InstallMeta, update, Result};
#[derive(Debug)]
pub struct Selection {
pub source: Source,
// Retained for the frontend response shape; direct updates never offer a bridge.
pub latest_version: Option<String>,
}
pub fn select(meta: &InstallMeta, source: Source, key: &str) -> Result<Selection> {
select_with(meta, source, &mut |source| pkgsource::fetch_manifest(source, key))
}
fn select_with(meta: &InstallMeta, source: Source, fetch: &mut impl FnMut(&Source) -> Result<(Manifest, Vec<u8>)>) -> Result<Selection> {
let (manifest, _) = fetch(&source)?;
update::validate_candidate(meta, &manifest)?;
Ok(Selection { source, latest_version: None })
}
#[cfg(test)]
mod tests {
use super::*;
use base64::{engine::general_purpose::STANDARD, Engine};
use ed25519_dalek::{Signer, SigningKey};
use std::{fs, path::Path};
fn package() -> serde_json::Value {
serde_json::json!({"schema":1,"kind":"update","version":"0.261015.58","core":{"commit":"a".repeat(40)},
"builtAt":"test","minManagerVersion":"0.6.16","files":[],"migrations":[],"archive":null,
"compatibility":{"schema":1,"platform":if cfg!(windows) {"windows-x86_64"} else {"linux-x86_64"},
"coreCommit":"a".repeat(40),"moduleCommits":{"squid":"b".repeat(40)},"clientPatchVersion":"1.7",
"sourceVersions":["0.261013.54"],"sourceDatabases":{"0.261013.54":{"manifestSha256":"c".repeat(64),"schemaSha256":null}}}})
}
fn meta(version: &str) -> InstallMeta {
let mut meta = InstallMeta::new(crate::registry::InstallKind::New, Path::new("unused"));
meta.core.version = Some(version.into());
meta
}
fn fixture(value: &serde_json::Value, tamper: bool) -> (tempfile::TempDir, String) {
let folder = tempfile::tempdir().unwrap();
let signing = SigningKey::from_bytes(&[37; 32]);
let bytes = serde_json::to_vec(value).unwrap();
fs::write(folder.path().join("manifest.json.sig"), STANDARD.encode(signing.sign(&bytes).to_bytes())).unwrap();
fs::write(folder.path().join("manifest.json"), if tamper { b"tampered".as_slice() } else { &bytes }).unwrap();
(folder, STANDARD.encode(signing.verifying_key().to_bytes()))
}
#[test]
fn every_known_older_version_selects_the_latest_signed_package_once() {
for installed in ["0.2.0", "0.261004.20", "0.261005.0", "0.261007.24", "0.261011.800006", "0.261013.54", "0.261015.58"] {
let (folder, key) = fixture(&package(), false);
for kind in [crate::registry::InstallKind::New, crate::registry::InstallKind::Imported] {
let mut installation = meta(installed);
installation.kind = kind;
let mut calls = 0;
let selected = select_with(&installation, Source::Dir(folder.path().into()), &mut |source| {
calls += 1;
pkgsource::fetch_manifest(source, &key)
}).unwrap();
assert!(matches!(selected.source, Source::Dir(path) if path == folder.path()));
assert!(selected.latest_version.is_none());
assert_eq!(calls, 1, "no intermediate package fetches for {installed}");
}
}
}
#[test]
fn unknown_versions_and_downgrades_remain_blocked() {
let (folder, key) = fixture(&package(), false);
for installed in ["unknown", "a65be06ec62f4", "0.261016.59"] {
assert!(select(&meta(installed), Source::Dir(folder.path().into()), &key).is_err());
}
}
#[test]
fn signed_target_is_required_even_for_unlisted_old_versions() {
let (folder, key) = fixture(&package(), true);
assert!(select(&meta("0.261004.20"), Source::Dir(folder.path().into()), &key).is_err());
}
#[test]
fn platform_and_manager_requirements_are_still_enforced() {
for (field, value) in [("minManagerVersion", "99.0.0"), ("platform", "other-platform")] {
let mut target = package();
if field == "platform" { target["compatibility"][field] = value.into(); }
else { target[field] = value.into(); }
let (folder, key) = fixture(&target, false);
assert!(select(&meta("0.261004.20"), Source::Dir(folder.path().into()), &key).is_err());
}
}
}
+25
View File
@@ -47,6 +47,31 @@ fn run() -> Result<(), String> {
let e = |x: coa_core::Error| x.to_string();
let part = a.get("part-size").and_then(|v| v.parse().ok()).unwrap_or(DEFAULT_PART_SIZE);
match cmd.as_str() {
"plan-update" => {
let mut meta = coa_core::registry::InstallMeta::new(coa_core::registry::InstallKind::New, std::path::Path::new("unused"));
meta.core.version = Some(need(&a, "from")?.clone());
let selection = coa_core::update_route::select(&meta, coa_core::pkgsource::Source::Url(need(&a, "source")?.clone()), coa_core::signing::EMBEDDED_PUBLIC_KEY).map_err(e)?;
let (manifest, _) = coa_core::pkgsource::fetch_manifest(&selection.source, coa_core::signing::EMBEDDED_PUBLIC_KEY).map_err(e)?;
println!("{}", serde_json::json!({"from":meta.core.version,"next":manifest.version,"latest":selection.latest_version}));
}
"verify-install" => {
coa_core::release_schema::verify_install_package(
&PathBuf::from(need(&a, "dir")?), coa_core::signing::EMBEDDED_PUBLIC_KEY,
).map_err(e)?;
}
"compose-install" => {
coa_core::release_schema::compose_install(
&PathBuf::from(need(&a, "base")?), &PathBuf::from(need(&a, "candidate")?),
&PathBuf::from(need(&a, "fixture")?), coa_core::signing::EMBEDDED_PUBLIC_KEY,
).map_err(e)?;
}
"qualify-install" => {
let report = coa_core::release_schema::qualify_install(
&PathBuf::from(need(&a, "package")?), &PathBuf::from(need(&a, "fixture")?),
coa_core::signing::EMBEDDED_PUBLIC_KEY,
).map_err(e)?;
coa_core::fsx::atomic_write_json(&PathBuf::from(need(&a, "report")?), &report).map_err(e)?;
}
"qualify-upgrade" => {
let report = coa_core::release_schema::qualify_upgrade(
&PathBuf::from(need(&a, "base")?), &PathBuf::from(need(&a, "source")?),
-655
View File
@@ -1,655 +0,0 @@
# CoA Server Manager — Phase 0 audit & architecture
Status: Phase 0 output. Everything under **Audit findings** was checked against the real repositories and a real
local install (`C:\games\CoA-Repack`); nothing there is invented. Everything under **Decisions** is a proposal
that the phases below implement.
---------------------------------------------------------------------------------------------------------
## 1. Audit findings (facts)
### 1.1 The repack already defines the on-disk layout
`C:\games\CoA-Repack` (RELEASE.json `mainRevision c3beca68`, `sourceRevision b8615469`, 2026-09-11):
```
Core\ authserver.exe worldserver.exe, DLLs, configs\, configs\modules\, Logs\, Crashes\
Data\ maps vmaps mmaps dbc (incl. dbc\Ascension) Cameras 3.8 GB
mysql\ MySQL 8.4.9 (mysqld, mysql, mysqladmin, mysqldump), data\ (3.3 GB), data.7z (141 MB)
Runtime\python\ embedded Python used by Scripts\
Scripts\manage.py the launcher/supervisor (604 lines) + gui_server.py (local web GUI)
Settings\ repack.json (ports, RA creds), database.json (generated DB passwords),
*.conf.template (+ .original)
BugReport\ relay.py, reports\, Logs\
Source\ server-source (2.2 GB) + zip + source-manifest.json
.state\ <name>.json = {pid, exe, created(FILETIME)} for mysql/auth/world/supervisor/relay,
control.lock, configuration.json, debug-logs.json
RELEASE.json {mainRevision, sourceRevision, binaries[{Name,SHA256,Bytes}], account, ...}
MANIFEST.json {fileCount 19382, bytes 3.2 GB, files{path:{bytes,sha256}}} <- per-file hashes
*.bat thin wrappers around Runtime\python\python.exe Scripts\manage.py <verb>
```
Consequences:
* **A per-file hash manifest already exists** (`MANIFEST.json`). It is the baseline for the file-ownership
system on imported repacks: file present in MANIFEST with the same hash → `core/repack, pristine`; present with a
different hash → `modified`; not in MANIFEST → `user`. No need to hash 3 GB to classify.
* **Generated configs are overwritten on every start.** `manage.py prepare()` renders
`Settings\*.template` → `Core\configs\{authserver,worldserver}.conf` and
`modules\mod_ascension_compat.conf` (placeholders `@DATA@ @LOGS@ @LoginDatabaseInfo@ ...`), also writes
`mysql\my.ini`, `mysql\admin-client.ini` (contains the root password) and `modules\coa_bugreport.conf`.
The README says so explicitly. **The templates, not the `.conf` files, are the source of truth** for those three
configs. Other module confs (e.g. `mod_coa_playerbots.conf`) are *not* templated and are edited in place.
* **Process identity is already solved in the repack** the way the spec demands: `{pid, exe, creation FILETIME}`
via `QueryFullProcessImageNameW` + `GetProcessTimes`; listeners via `GetExtendedTcpTable` with owning PID;
a start/stop mutex via `msvcrt.locking` on `.state\control.lock` (1 byte at offset 0 — compatible with
`LockFileEx`); graceful stop = RA `server shutdown 1s 0`, then auth terminate, then `mysqladmin shutdown`.
`stop_all` never hard-kills worldserver. Port offset support exists (`--port-offset`).
* **World is started under a supervisor** (`watch-world`) that also runs the Python bug-report relay; starting
`worldserver.exe` directly skips the relay. The relay ships reports to the maintainers.
* **Ports:** MySQL 3307 (bind 127.0.0.1), auth 3724, world 8085, RA 3443 (RA user/pass in `repack.json`).
* **DB credentials are the *same for every repack download*** (`database.json` is packaged, 48-hex passwords).
Safe only because MySQL is loopback-only. New Manager-created installs must regenerate them.
* **Readiness marker** in `Core\Logs\Server.log`:
`AzerothCore rev. <12-hex> <date> (main branch) (...) (worldserver-daemon) ready...` and
`WORLD: World Initialized In N Minutes M Seconds`. The worldserver banner also **reports its own core commit**
(12 hex) — usable to identify the core version of an imported binary when RELEASE.json is stale.
### 1.2 The reference machine is a real, customised, *running* install
* `Core\worldserver.exe` sha256 `b9aa8f36…` ≠ RELEASE.json (`715bd375…`): the owner's dev build. `Core\`
holds ~60 `worldserver.exe.pre-*` backups and `.pdb`s — **157 GB**. Import must not walk-and-hash everything;
it hashes only what it needs (exes, configs, templates, addon) and enumerates the rest by name/size.
* `mod_coa_playerbots.conf` has 71 `CoaBots.*` keys while the repo's `dist/conf/*.dist` has ~30: real configs
are ahead of the shipped template. The bot schema must therefore tolerate unknown keys (spec §10 already
demands it) and be validated against the *deployed* `.dist`, not only the repo's.
* The worldserver on this machine is **running now (pid 22080)** and the owner runs long bot experiments on it.
**Testing rule: the live install is only ever read.** Start/stop/update tests run on a disposable fixture copy
on different ports (`--port-offset`), never on `C:\games\CoA-Repack`.
* `C:` has 86 GB free (94 % full). Backup/staging code must check free space *before* writing and default
large artefacts (staging, DB dumps) to a user-selectable location.
### 1.3 mod-coa-playerbots is not a plug-in — it is a rebuilt worldserver
`Corfirean/mod-coa-playerbots` (`master`, no GitHub Releases yet) ships `dist\`:
```
bin\worldserver.exe (not in git) Install-CoaBots.ps1/.bat Uninstall-CoaBots.ps1/.bat
conf\mod_coa_playerbots.conf.dist reference\ascensionsidekick-level-builds.json (282 KB)
addon\CoABotUI\{CoABotUI.lua,.toc} sql\2026_09_15_00_ascension_character_selection.sql
release.json {builtForRepack{mainRevision, sourceRevision}, originalWorldserverSha256[], patchedWorldserverSha256[]}
```
* The module is compiled **statically into `worldserver.exe`** (`-DMODULES=static`) *and* needs three small
**core patches** (`docs/core-patches.md`: `LoginQueryHolder` moved to `WorldSession.h`,
`IWorld::AddQueryHolderCallback`, `Group::GetRolls`, plus an Ascension talent accessor). The doc says some of
these were still uncommitted in the core working tree — **must be verified before CI can build bots
from clean commits** (§6 risk R1).
* "Install bots" today = replace `Core\worldserver.exe` (backup → `.orig`), drop `.conf.dist`, create `.conf`
only if absent, copy `Core\reference\ascensionsidekick-level-builds.json`, copy the addon to
`<repack>\Client\Interface\AddOns\CoABotUI` (a `Client\` folder that the repack does not have — the client is
distributed separately), apply `dist\sql\*.sql` to the characters DB with the password read out of
`worldserver.conf` (visible on the command line).
* The installer already does **hash-gated binary replacement** (original ∈ known hashes or patched ∈ known
hashes, else refuse unless `-Force`) and refuses if worldserver is running. The Manager generalises this.
* Bots are real characters on their own accounts (`CoaBotHost*`) in `acore_characters`; no extra database.
Bot creation throttles (`BatchSize 5 / 500 ms`) exist because a 1000-bot synchronous spawn once killed the
server — presets must never lower these. (Confirmed by the owner's own incident notes.)
* The `.conf.dist` is well commented prose — usable as the source for the initial `schemas/bots.json`.
### 1.4 azerothcore-wotlk-coa (core)
* `main`, 1.1 GB, **no releases, no tags, no Windows build workflow**. Only `quality.yml` (ubuntu, source
boundaries) exists. CI for binaries must be written from scratch.
* Build: CMake ≥ out-of-source, `-DSCRIPTS=static -DMODULES=static`, C++20, Boost.PropertyTree (vcpkg
`boost-property-tree`), OpenSSL 3, MySQL client libs. Repo rules (`AGENTS.md`): never edit
`data/sql/base|archive|updates/db_*`; new SQL goes only in `data/sql/updates/pending_db_*/`; **automatic DB
updater is disabled in this project's config** (the bots doc: `AUTOUPDATER: ... disabled`). So schema
migrations are *not* applied by worldserver — the Manager must own a tracked migration runner (spec §18).
* `modules\` contains 24 in-tree modules (mod-ascension, mod-coa-*, …); bots would be the 25th.
Runtime module configs are `Core\configs\modules\*.conf` (+ `.dist`).
* Core HEAD 442cf4c9 (2026-09-30) is ahead of the repack's `mainRevision c3beca68` and the running binary
(`3567e2f8e9d5`): three different revisions are in play on one machine. The compatibility matrix must be
keyed by **exact core commit**, never by "latest".
### 1.5 Toolchain on this machine
Node 26, npm 11, git 2.55, gh (authed as Corfirean), VS 18 Insiders with MSVC 14.51 and WebView2 present.
Rust was **not** installed; `rustup` stable 1.98.1 (msvc) was installed as part of this task. cmake/pnpm absent
(npm is used; CMake is only needed by the *build* repo, on the CI runner).
---------------------------------------------------------------------------------------------------------
## 2. Decisions (architecture)
### D1. One supported layout: the "CoA Repack layout v1"
Both flows converge on the same folder shape:
* **Import** adopts an existing repack-shaped folder (classification: Healthy / Partial / Unknown-custom /
Incompatible, §5).
* **New install** downloads the *base package* (= a repack) and extracts it into staging, verifies, then moves
it into place. Every install is therefore a repack plus Manager metadata.
Result: one code path for process control, config, DB and update. "Unknown custom server" gets the restricted
mode from spec §52 (start/stop via detected exes, backups, console, no update/bots).
A `Layout` trait isolates this (`RepackLayout` now; a hypothetical `NativeLayout` later) so nothing hardcodes
`Core\worldserver.exe`; paths come from `Installation.layout`.
### D2. Manager metadata lives *next to*, not inside, the server folder
`D:\CoA-Repack` ⇒ `D:\CoA-Repack.manager\` (the example in the spec). Rationale: import stays byte-for-byte
read-only on the server folder, folder can be copied/moved without dragging Manager state, backups get a
natural home the user can redirect to another drive.
```
<install>.manager\
install.json identity + detected versions + db info (no secrets) + client path
state.json last known process state, update-transaction status pointer
ownership.sqlite (or ownership.json in v1) path → {owner, hash, version, replacePolicy, origin}
manifests\ copies of every applied release manifest
backups\<id>\ recovery points (each has backup.json + payload)
migrations\ applied-migration ledger (mirrors the DB table, see D7)
staging\ update/install staging (same volume as install → atomic rename)
logs\ manager.log (rotated 5×2 MB) ; cache\
```
A registry at `%LOCALAPPDATA%\CoAServerManager\installs.json` maps `installationId → path`, so a *moved* folder
is re-linkable (the `.manager` folder holds the id; if the path is gone the UI offers "Locate server").
Secrets (DB/RA passwords generated by the Manager) go to **Windows DPAPI-protected** files in the same
`.manager` folder (`secrets.bin`, `CryptProtectData`, user scope). Imported servers keep their own
`Settings\database.json`; the Manager reads it in memory, never displays or logs it.
### D3. Process model (Rust, `process` module)
* `ProcessIdentity {pid, exe_path, creation_time_filetime}` — identical to `.state\*.json`, so records are
interchangeable with the repack's own scripts.
* A process "belongs" to an installation iff `exe_path` is under the installation root (or equals a
configured exe) **and** creation time matches a recorded identity *or* it owns a listening socket on a
configured port. Never by image name alone; never by bare PID.
* Discovery on startup = scan running processes for the installation's exes + `GetExtendedTcpTable` owner PIDs →
rebuild state (works after a Manager crash, PC crash → stale PID dropped because creation-time/exe mismatch).
* Start/stop are serialised with the repack's `.state\control.lock` via `LockFileEx` (same byte), so Manager and
`.bat` files cannot race.
* **Driver strategy.** Phase 2 delegates start/stop to the repack's own `manage.py` through its bundled Python
(`Runtime\python\python.exe -B Scripts\manage.py start-all|stop-all`), because it already handles config
rendering, relay, RA shutdown and readiness, and because behaviour must not regress on a production install.
Observation (identity, ports, health, logs) is native from day one. A native Rust driver for the same
sequence replaces the delegate later, gated behind the same `ServerDriver` trait and validated against the
delegate by the same integration tests. Startup order/health follows spec §7–8.
* States: `Stopped | Starting | Running | Stopping | Crashed | Updating | Unknown`; overall health adds
`Ready | NeedsAttention` using: process alive → port owned by that PID → log `ready...` marker after this
start's timestamp → no `FATAL`/`ERROR` startup lines.
* Forced kill exists only as an explicit, warned last resort after graceful stop times out.
### D4. Config engine (`config` module)
* Line-preserving parser: keeps comments, blank lines, unknown keys, line endings and encoding (UTF-8, LF/CRLF
detected). A `set(key, value)` rewrites only that key's line; missing keys are appended under a
`# Added by CoA Server Manager` comment. Never re-serialises the file.
* Layered targets via the layout: for templated configs the edit is applied to `Settings\<name>.template`
(source of truth) **and** the rendered `.conf` (so the running/next state agrees without a restart of the
renderer); for `mod_coa_playerbots.conf` edit in place.
* Save = validate against schema → write `<file>.tmp` in same dir → re-parse and compare → snapshot previous
into `.manager\backups\config\` → `MoveFileEx(REPLACE_EXISTING)`.
* Merge (`.dist` v1 → v2): add keys present in new `.dist` but missing; never overwrite user values; keys
removed upstream → flagged `deprecated`, kept.
* Schemas: `schemas/bots.json`, `schemas/server.json` (curated whitelist), fields exactly per spec §10
(`key type category title description default min max options advanced restartRequired dangerous`;
`restartRequired ∈ runtime|world|full`). Presets are separate JSON diffs shown as "will change N settings".
### D5. File ownership & manifests
Every release ships `manifest.json`:
```jsonc
{
"schema": 1,
"kind": "base|update|bots",
"version": "0.4.1",
"core": { "commit": "<40 hex>" },
"bots": { "commit": "<40 hex>" | null },
"builtAt": "2026-…Z",
"minManagerVersion": "0.1.0",
"requires": { "core": ["<commit>"|"range"], "dbSchema": ">=N" },
"files": [ { "path":"Core/worldserver.exe", "sha256":"…", "size":123, "owner":"core|bots|manager",
"policy":"replace|replace-if-pristine|merge-config|create-if-missing|never-touch" } ],
"migrations": [ { "id":"2026_09_15_00_…", "db":"characters", "sha256":"…", "destructive":false } ],
"signature": { "alg":"ed25519", "keyId":"…", "sig":"…" } // verified when a public key is embedded (§54)
}
```
`ownership` table per install: `path, owner(core|bots|manager|user), sha256_at_install, installed_version,
policy, origin(imported|installed)`. Update algorithm compares **current hash vs recorded hash vs incoming
hash**: current == recorded → safe to replace; current ≠ recorded → *modified outside Manager* → prompt
(Keep / Replace / Compare / Cancel); path unknown → `user`, never touched. Import seeds the table from
`MANIFEST.json` where present (imports are read-only; hashes computed lazily and only for managed paths).
### D6. Update transaction (Prepare → Verify → Snapshot → Apply → Validate → Commit)
1. **Prepare** — fetch manifest (HTTPS only, signature checked), compatibility matrix check, disk-space check on
staging volume and backup volume.
2. **Download** to `staging\<txid>\dl\` with `.part` files, HTTP Range resume, retry/backoff, cancel, progress.
3. **Verify** — SHA-256 of every artefact; nothing is executed or extracted into the live tree before this.
4. **Snapshot** — server must be stopped (never copy over a running worldserver); create recovery point:
replaced binaries, changed managed configs, module state, full dumps of every managed realm database;
write `txn.json {state: Snapshotted}`.
5. **Apply** into `staging\<txid>\tree\`, resolve ownership conflicts, merge configs, then swap file-by-file with
same-volume atomic renames, journaling each step (`txn.json` lists done/undone operations).
6. **Migrate** — durably mark SQL as running before execution. Failure restores the full database recovery point
before restoring files. Interrupted SQL is never automatically replayed.
7. **Validate** — check database structure, start server, require ten seconds of continuous readiness with timeout.
8. **Commit** — ownership table + `install.json` version bump, mark txn `Committed`, prune staging.
Any crash before Commit leaves `txn.json` in a non-terminal state; on next launch the Manager detects it and
offers *Roll forward* or *Roll back* — the install is "recoverable" by construction.
Rollback restores files, configuration and installation metadata. Once migrations started, it also restores all
databases from the full pre-update recovery point. Failed recovery remains unfinished and blocks startup and another
update. The same journal protects Repair. Old transactions without a full recovery point cannot roll back database
changes automatically. Independent database restore remains available for other recovery points.
### D7. Database strategy (`db` module)
* Driver: talk to the repack's own `mysql.exe` / `mysqldump.exe` / `mysqladmin.exe` (bundled, 8.4.9). Passwords
are passed via a temp `--defaults-extra-file` with restrictive ACL, deleted immediately — never on argv. A
native Rust client (`mysql_async`) is used for health pings and migration ledger reads.
* Ledger: table `acore_world.coa_manager_migrations(id, db, sha256, applied_at, status, error)` (created by the
Manager, in a dedicated table — no core schema is changed) mirrored to `.manager\migrations\`.
Statuses `Applied | Pending | Failed | Running`. Each file runs in a single connection; DDL is not transactional, so
"destructive" migrations (flag in manifest, or heuristic scan for `DROP|TRUNCATE|DELETE|ALTER ... DROP`) force
a fresh DB backup first and require explicit confirmation.
* Backups: `mysqldump --single-transaction --routines --quick` per DB. Quick = characters + auth + configs +
metadata; Full adds world (large); Config-only; DB-only. Compressed with zstd. Restore is always to a
staging schema first (`…_restore`), swapped only after a row-count sanity check, with the current DB dumped
beforehand.
* Never: `DROP DATABASE` or resetting characters. Update recovery restores databases into staging schemas and
keeps the replaced tables in separate schemas. Delete/
reset lives only in **Settings → Advanced → Danger Zone**, deletes tracked files only, and refuses to remove a
directory that still contains untracked files.
* Clean install: extract `mysql\data.7z` (packaged data dir), then **rotate** root/app passwords to fresh random
values (`ALTER USER`), rewrite `Settings\database.json`/templates via the layout renderer, store copies in
DPAPI. `bind-address=127.0.0.1` enforced.
### D8. Bot install/update (needs server-side cooperation — see §4)
Compatibility matrix `compat.json` (published with releases): `[{manager, core, bots, dbSchema}]` keyed by exact
core commit. Bots install = "install the bot-enabled `worldserver.exe` built for **this** core commit" +
conf.dist + talent JSON + SQL migration + (if client attached) addon. Same hash-gated + backup-first rules as
the existing `Install-CoaBots.ps1`, executed through the D6 transaction. Force-install on unknown core =
Advanced override, snapshot first.
### D9. Client integration
The client is a separate object. By default the Manager only detects it (`Wow.exe`/`Ascension.exe` + `Data\`), reads/writes
`realmlist.wtf` after backing it up, and copies `Interface\AddOns\CoABotUI` (backing that one directory up only if
its files differ from the manifest). No touching of `WTF`, `Cache`, other AddOns, or the exe. Downloading or updating
a client is opt-in and described in section 9 (game client download); an existing client is never changed without asking.
*Reference machine note:* `C:\games\Ascension` is a heavily modified Ascension client (custom `d3d9.dll`,
renderer files, many `.bak`s) — client detection must be strictly read-only + additive.
### D10. Network / friends
Detect LAN IP and public IP (HTTPS to a small set of IP-echo endpoints, user-toggle), listening check via
outbound self-test, **CGNAT heuristic** (WAN address of router ∈ 100.64.0.0/10 or private / ≠ public IP),
UPnP/NAT-PMP later. Firewall rules created only through a separate elevated helper (UAC on demand, never whole
app as admin), fixed names `CoA Server Manager - Auth` / `- World`, idempotent (query before add), delete only
own rules. Never exposed: MySQL 3307, RA 3443, Manager control channel (Tauri IPC only, no listener). Private
network = Tailscale detection/guided enable in v1; interface `NetworkProvider` so "CoA Connect" can slot in.
Friend package contains only `realmlist`/instructions/optional addon; secrets structurally excluded (whitelist
copy, not blacklist).
### D11. Stack & code layout
```
coa-server-manager/
docs/ ARCHITECTURE.md, flows, wireframes
schemas/ bots.json, server.json, presets/, manifest.schema.json, compat.schema.json
src-tauri/src/
main.rs, commands/ thin Tauri command layer (validate → call service)
fsx/ path validation, safe join, atomic write/replace, disk space, ownership
layout/ RepackLayout, detection & classification
registry/ installs.json, .manager dir, install.json
process/ identity, listeners, lock, drivers (delegate → native)
health/ probes, log-marker scan, error translation
config/ parser, merge, schema, presets
db/ mysql tool wrapper, ledger, migrations, dump/restore
backup/ recovery points, retention
update/ manifest, download, verify, txn (journal), rollback
client/ detection, realmlist, addon
net/ ips, ports, cgnat, firewall, tailscale
errors/ error catalogue → human message + actions
diag/ diagnostics, export package (redacting)
src/ React + TS + Vite + Tailwind + shadcn/ui (presentation only)
tests/ integration fixtures (fake repack generator), failure-injection harness
coa-server-build/ (separate repo) workflows + packaging + manifest signing
```
All destructive filesystem operations exist only in `fsx`; frontend never has `fs` scope in Tauri capabilities.
Every command that mutates goes through `fsx::Transaction`-style helpers that (a) validate the path is inside a
registered installation *or* the `.manager` dir, (b) refuse symlink/junction escapes, (c) log an audit record.
### D12. Build/release repo (`coa-server-build`)
Windows runner: checkout **exact** core SHA and bots SHA from workflow inputs (refuse branch names), place bots at
`modules/mod-coa-playerbots`, verify core patches are already in the core commit (fail the build if the patch set
is not applied — never apply ad-hoc patches at build time unless they are versioned `.patch` files in the build
repo), vcpkg-cached deps, CMake Release/RelWithDebInfo static, produce **two binaries per core commit**:
`worldserver` (no bots) and `worldserver+bots`, then `authserver`, DLLs, config `.dist`s, talent JSON, addon, SQL
migrations, `RELEASE.json`, `manifest.json` (+SHA256, +ed25519 signature), zstd-compressed update package.
Base package (≈3.2 GB; Data 3.8 GB + mysql data) is split into ≤ 1.9 GB parts (GitHub release asset cap is 2 GiB)
with a parts manifest. Reproducibility: pinned runner image label, recorded compiler/vcpkg versions, source SHAs,
build timestamp written into RELEASE.json.
### D13. Linux: the servers run in Docker (experimental)
Linux has no repack launcher and no Win32 process API, and the host should need nothing but Docker. A server folder
that contains `Settings/docker.json` is a **Docker installation**; every other folder is a repack, handled by the
code above without any change. `driver::run`, `process::observe` and `layout::scan` check for the marker first and
hand over to `coa_core::docker`; nothing else needed to change for start, stop, status and the server list.
* **Containers** (`docker/lifecycle.rs`): `coa-<project>-db` (official `mysql:8.4`, data in a named volume, not
published on the host), `-world` and `-auth` (a small runtime image holding only the shared libraries; the
binaries are mounted from `Core/`, so an update replaces files and never rebuilds an image). The image tag is a
hash of its Dockerfile.
* **Why the `docker` command line and not `docker compose`**: the compose plugin is missing from some distributions'
Docker packages, and the plan is to ask for as little as possible on the host. The command line is wrapped by a
`Docker` trait (`docker/cli.rs`), so the logic is tested with an in-memory fake.
* **Same folder shape as a repack where it matters**: `Core/configs`, `Core/Logs`, `Settings/repack.json`
(ports, RA login), `Settings/database.json` and `Data/`. Configuration editing, log tailing and the RA console
therefore work as they do on Windows. The container runs as the owner of the folder and starts in `Core/`.
* **Settings forced by environment** (the core reads `AC_<KEY>` before the `.conf`): `AC_BIND_IP=0.0.0.0` (the
published address, `bindAddress`, decides who can connect), `AC_DATA_DIR`, `AC_LOGS_DIR`, `AC_RA_ENABLE`,
`AC_RA_IP`, and `AC_UPDATES_ENABLE_DATABASES=0`: the server's own database updater needs a `mysql` client program and
stops the server without one; the Manager applies database updates itself (section 1.4). The remote console is published
on `127.0.0.1` only.
* **Secrets** go to `docker` by environment (`--env NAME` without a value), never on a command line.
* **Order**: database healthy, then world listening, then auth. The auth server exits when no realm is online.
Readiness is a connection that stays open: Docker's proxy accepts connections on a published port before the server
inside listens, so a plain connect proves nothing.
* **Configuration path in the binaries**: Linux builds must use `-DCONF_DIR=configs` (see `coa-server-build`),
otherwise the core looks for module configs in a path baked in at build time.
* **Database access** (`db.rs`): for a Docker installation `Db` runs `mysql`, `mysqldump` and `mysqladmin` inside the
database container with `docker exec -i`, connecting over TCP on the container's loopback exactly as the repack's
commands do on the host (password through `MYSQL_PWD`, never on a command line). Nothing else changes: accounts,
backups, the population query, migrations and the RA service account all go through `Db`, and the host needs no
MySQL client. A repack keeps using its bundled tools.
* **Health of the database container** is a ping over TCP, like every connection the Manager makes. On a new volume the
image first runs a temporary server that answers on its socket but takes no network connection; a socket ping reported
the container healthy about four seconds before TCP worked (measured: 10 s against 14 s).
* **Main configuration files**: a package carries only `worldserver.conf.dist` and `authserver.conf.dist` (the release
tool leaves active files out). On Windows the launcher writes them at each start; on Docker `ensure_main_configs` creates
them from the templates when they are missing, at installation and at every start, and never overwrites one.
* **Installing a new server** (`docker/install.rs`, separate from `install::install_base`, which is not touched). Same
order of events: signed manifest, space and folder checks, verified download, extraction into `<dest>.installing`,
database, then one rename. The package must hold `Core/worldserver`, `Core/authserver`, the `.dist` templates,
`_migrations/` and **`Database/baseline/{auth,characters,world}.sql.zst`**. The baseline is a dump, in the format of the
Manager's backups, of databases in which **every migration of the package is already applied**; the installer imports
it, records those migrations as applied (`baseline = 1`, a few bulk statements) and creates the `acore` account, the
realm port and the console account. It is the same arrangement as a Windows package, which ships its prepared database.
The databases are **not rebuilt from the repository's SQL**: some migrations are guards generated from the maintainers'
live database (`rev_20260903_01_live_class_baseline` refuses to run unless the table holds exactly the rows it expects),
so replaying the SQL files from the base scripts fails, in any order. `coa-release export-baseline --server DIR --out DIR`
produces the dumps from a prepared server (repack or Docker). The game data stays where the person has it
(`dataDir` in `docker.json`, mounted read-only). A failure removes the containers, the volume and the staging folder.
* **Validating a release on Linux** (`docker/fixture.rs`, used by `coa-release extract-schema-base` and `schema-contract` where
there is no Windows database program): the signed base package carries the MySQL data directory of a Windows repack, and a
MySQL 8.4 container opens it as it is (`--lower-case-table-names=1`, the repack's setting). The fixture is a Docker server
folder (`docker.json` with `mysqlData`: the database container uses that directory in place of a volume; an installation
never sets it), plus a `root` account for TCP, which the repack only has for its socket. The release SQL is then applied and the
schema contract captured by the same code as on Windows, and `export-baseline` dumps the result: that is how a Linux build
produces `Database/baseline` without anyone's live database. The migration checksums recorded in the base are those of
files checked out on Windows (CRLF line endings, `core.autocrlf`): the SQL of the core, the modules and SQUID must be
checked out the same way for the checksums to match, or every applied migration looks edited.
* **In the screens** the host is asked once per command (`coa_core::platform::flavor()`: repack on Windows, Docker
elsewhere) and the Docker half lives in `src-tauri/src/docker_install.rs`; the repack arm of each install command is the
code that was there. `install_environment` tells the install screen the kind of server, the suggested folder
(`~/CoaServer`) and why Docker cannot be used, if it cannot. On Docker the screen also asks for the game data folder and
the package, and there is no default package address until a signed Linux package is published. Packages are checked with
the embedded key; only a debug build (`tauri dev`) accepts another one, named in `COA_DEV_TRUSTED_KEY`, to try the
installer on a package signed with a throwaway key.
* **Updates are refused for a Docker server, never half done.** `update::preview`, `apply` and `rollback` refuse it: the
update packages are the repack's (executables, a bundled MySQL, a launcher) and applying one would put Windows files in a
Linux folder. Found by installing a server with the screen: the update check had read the Windows channel and offered one.
* **Realm profiles (CoA / Wildcard) work on a Docker server.** What the repack's patched launcher did at each start is done
by the Docker backend: finish an interrupted switch (`realms::before_start`), name the realms in the realm list, and give
the containers the databases of the selected realm (`acore_world_wildcard` ...). The switch itself (configuration swap,
database copy, journal) is the shared code; `prepare_launcher` has nothing to patch, and Wildcard support is read from
`Core/worldserver` as it is from `Core/worldserver.exe`.
* **The game client starts under Wine on a Linux host** (`wine.rs`; `client::launch` and `client::is_running` hand over):
umu-launcher with a Proton build when `umu-run` is installed (the newest GE-Proton found for Steam, or umu's own choice),
else plain Wine; `WINEPREFIX` defaults to `~/Games/umu/coa-client`; `WINEDLLOVERRIDES=divxtac=d` is always set, because
DivxTac.dll deadlocks the world loading screen under wine-mono. `COA_PROTONPATH`, `COA_WINEPREFIX` or
`client-launch.json` in the Manager's folder (`proton_path`, `prefix`, `runner`) override the detection. Whether the client
runs is read from `/proc` (under Wine the command line holds the Windows form of the path). Output goes to `client.log`.
* **Not done yet**: downloading the game data, updates, port conflict detection, the firewall and exposure checks, the
Wildcard realm profiles, and the client under Wine/Proton.
### D14. Keeping Windows and Linux apart
A fix for one platform must not be able to break the other. The rules:
1. **Platform code lives in its own modules**: `coa_core::docker` for Linux and Docker; the repack code (`driver`, `process`,
`install`, `layout` and the bundled-tool half of `db`) for Windows. A new Linux feature is a new file there, not an edit
of a repack function.
2. **Shared modules hold platform-neutral logic only** (manifest, signing, download, extraction, migrations, backups,
configuration editing, the registry).
3. **The difference enters in as few places as possible**: the early `is_docker` return at the top of `driver::run`,
`process::observe`, `layout::scan` and `Db::from_repack`, and the platform checks of the Tauri layer. Everything past
that point belongs to one platform.
4. **No `cfg` in the middle of a shared algorithm.** Where one is unavoidable (file modes, a path comparison), it is small,
commented, and the other platform's branch is the code that was there before.
5. **Reuse by calling, not by editing**: a Linux feature that needs a helper from a Windows-side file makes it
`pub(crate)` and calls it (visibility only); it does not change what the helper does.
6. **Both platforms are tested on every pull request** (`check.yml`: Windows and Linux). Tests of one platform's behaviour
are marked for that platform, so they cannot fail on the other.
---------------------------------------------------------------------------------------------------------
## 3. Flows
### 3.1 First run
`Welcome → [Install new server] | [I already have a server] | (link) Manage another install`.
New: choose folder (default `C:\Games\CoA Server`; checks: free space ≥ base+backup headroom, folder empty or
absent, not a client folder, not `Documents`/system dirs, not already registered) → single progress screen
(Download → Verify → Extract → Database → Bots(optional) → Start → Health) → "Your server is ready" →
Create account → PLAY. Technical log behind "Show details" only.
### 3.2 Existing-server import (strictly read-only)
select folder → `layout::scan()` (read-only opens, `FILE_SHARE_READ|WRITE` so a live server is unaffected) →
report: core exes + hashes vs RELEASE.json/MANIFEST.json, detected core commit (RELEASE.json → else banner rev in
Server.log → else unknown), configs & modules present, DB running/reachable (ping only), bots detected (conf
present ∧ patched-hash or `CoaBots.*` keys), client detected, classification + "No files will be changed" →
**[Add server]** writes only `<folder>.manager\` and the registry entry. Import never "fixes" anything.
Verification for the fixture test: recursive (path,size,mtime) snapshot of the server folder excluding
files the running server itself changes, diffed before/after.
### 3.3 Update — see D6. Failure matrix (spec §60) is executable via a fault-injection harness: kill at step k,
disk-full at write n, network cut at byte n, migration returns error, new worldserver exits non-zero, port
taken, exe quarantined (file vanishes), folder moved, user-modified managed file, unknown config keys.
### 3.4 Rollback — restore the pre-update databases when migrations started, followed by files and metadata
The UI explains that progress after the recovery point is lost. SQL failures before startup recover automatically;
failed health checks keep the installation blocked until the owner chooses recovery. Every recovery point and staging
schema has a unique identifier, and corrupt update journals fail closed.
---------------------------------------------------------------------------------------------------------
## 4. Required changes in the two source repositories
Prefer server-side cleanliness over Manager hacks. Proposed (each is an issue/PR on the owner's side, none is
made without approval):
| # | Repo | Change | Why |
|---|------|--------|-----|
| S1 | mod-coa-playerbots | Master switch `CoaBots.Enable` (0 = module registered but inert; no login/hooks/commands) | Lets a **single** worldserver ship for everyone. "Disable bots" becomes a config flip with restart, not a binary swap; drops the 2-binary matrix and most of §D8 risk. |
| S2 | mod-coa-playerbots | Commit **all** core patches in the core repo; delete "uncommitted working-tree" state from docs | CI must build from clean SHAs (R1). |
| S3 | mod-coa-playerbots | Real Releases + `dist/release.json` replaced by the shared `manifest.json` schema; `TalentBuildsPath` default = `Core/reference/…`, remove the hard-coded dev path fallback | Manager and installer share one truth; no absolute dev paths. |
| S4 | mod-coa-playerbots | Ship a machine-readable config schema (`conf/…schema.json`) next to `.dist`, or generate it in CI from the `.dist` comments | Avoid schema drift (71 vs 30 keys today). |
| S5 | core | Windows build+release workflow (or the build repo builds it); publish `worldserver --version --json` (core commit, bots commit, DB schema id) | Reliable version detection instead of log scraping. |
| S6 | core | Health endpoint alternative: worldserver already prints `ready...`; add a stable machine line (`COA-READY {...}`) or a tiny loopback status file `Core\Logs\ready.json` | Health without regex on human text. |
| S7 | core/repack | Ship a `manager-hooks.json` in the repack describing service names, ports, exes, config templates, RA settings | Layout detection without hard-coded knowledge; unknown custom builds can opt in. |
| S8 | core | Keep the auto DB updater disabled but expose `--dry-run-updates` listing pending updates | Manager migration UI without re-implementing the updater. |
| S9 | repack | Per-download random DB passwords and `Settings\database.json` generated on first start, not packaged | Security (§1.1). Manager does it itself for new installs meanwhile. |
Everything still works without S1–S9 (Manager has fallbacks), they just remove fragility.
---------------------------------------------------------------------------------------------------------
## 5. Import classification rules
* **Healthy** — `Core\{worldserver,authserver}.exe`, `Core\configs\*.conf`, `Data\dbc`, `mysql\bin\mysqld.exe` +
`mysql\data\acore_{auth,characters,world}`, `Settings\repack.json`, `Scripts\manage.py`, all present.
* **Partial** — a repack skeleton with pieces missing (e.g. no Data\maps, DB dirs absent).
* **Unknown/custom** — has `worldserver.exe` + `authserver.exe` but not the repack scaffolding.
* **Incompatible** — not a 3.3.5 AzerothCore layout / two installations nested / client folder.
The reference install classifies as **Healthy (customised)**: repack shape, binary hash differs from RELEASE.json.
---------------------------------------------------------------------------------------------------------
## 6. Risks & open questions
* **R1 (blocking for Phase 7/6)** — are all core patches committed? Owner action: confirm/commit (S2).
* **R2** — Legal/hosting of the base package: `Data\` (maps/vmaps/mmaps/dbc) derives from the game client;
and Ascension-specific DBC. Where does the owner want to host the 3–4 GB base (GitHub Releases in the build repo,
parts ≤ 1.9 GB; or own storage)? The Manager's downloader is source-agnostic; the choice affects Phase 5–6.
* **R3** — Manager signing key: needs an Ed25519 key pair generated by the owner; public key embedded in the app,
private key in GitHub Actions secrets. Phase 6 can start with unsigned + hash-only and flip the enforcement flag.
* **R4** — Bug-report relay (Python, sends reports to maintainers) is part of the repack's world startup; Manager
keeps it (privacy note surfaced in Settings, user can disable via the existing `coa_bugreport.conf` mechanism
only if the repack supports it — currently it forces `Enable = 1` on every start; not touched by Manager).
* **R5** — Tauri updater needs a hosted, signed `latest.json`; Manager self-update is Phase 10.
* **R6** — Windows long paths / non-ASCII install paths: repack requires a Latin path (uses 8.3 short names);
new-install folder chooser enforces it.
---------------------------------------------------------------------------------------------------------
## 7. Phase plan (each phase ends: compile → tests → run against fixture → verify live install untouched → commit)
| Phase | Deliverable | Done when |
|-------|-------------|-----------|
| 1 | Tauri+Rust+React skeleton, `fsx`, registry, `.manager` layout, logging w/ rotation, manifest structs + JSON schema, error catalogue skeleton | `cargo test` green; app launches; metadata round-trips |
| 2 | Read-only scan/classify/import; process discovery; Start/Stop via delegate driver; health | Import of `C:\games\CoA-Repack` leaves it byte-identical (proved by snapshot diff); Start/Stop verified on **fixture copy, offset ports** |
| 3 | Config parser/merge/atomic save; bots+server schemas; presets; UI | Property tests: unknown keys/comments preserved; round-trip identity |
| 4 | Recovery points, DB dumps, config snapshots, restore-to-staging | Backup→restore fixture test; fails safe when disk full |
| 5 | Clean install from base package; DB bootstrap + password rotation; account creation | Fresh VM/folder install reaches "ready" |
| 6 | Build repo & workflows; signed manifests; update transaction, rollback | Fault-injection suite green (spec §60) |
| 7 | Bots install/update via txn; compat matrix; GUI settings | Install → disable → restore on fixture |
| 8 | Client detection, addon, realmlist, PLAY | Client tree hash unchanged except addon dir + realmlist |
| 9 | Friends: net diagnostics, firewall service, Tailscale | Idempotent rule add/remove test |
| 10 | Error translation, diagnostics, repair, a11y, self-update | Spec §61 checklist |
---------------------------------------------------------------------------------------------------------
## 8. Decisions recorded after owner review (2026-09-30)
* **Core changes go through our own fork.** We are not maintainers of `jealous-sound/azerothcore-wotlk-coa`, so
the bot-enabling core patches (and the S1/S5/S6/S7 style changes) live in a fork owned by the bot/manager
project; the build repo pins exact fork SHAs. Upstream is tracked as a remote and merged periodically; the
compatibility matrix is keyed by fork commit. The fork is created when Phase 6/7 needs it (nothing is pushed
before then). This supersedes S2 (no upstream commits required) and removes risk R1.
* **Base package is compressed and split.** Payloads are zstd/LZMA-compressed and cut into parts below the
2 GiB GitHub asset limit (target 1.9 GB) with a parts list in the manifest; the downloader verifies each part
and the reassembled archive. Actual ratios will be measured in Phase 5 (the packaged database already shrinks
3.3 GB -> 141 MB; `Data\` maps/vmaps/mmaps are expected to compress far less).
* **Manifest signing (R3 resolved).** Ed25519 key generated. Public key: `keys/manifest-signing.pub`, embedded
in the app (`signing::EMBEDDED_PUBLIC_KEY`). Private key: `%USERPROFILE%\.coa-manager\signing\` (outside the
repo, never committed); CI receives it as the `COA_SIGNING_KEY` secret, which the owner must add manually.
Signatures are detached (`manifest.json.sig`, over the exact manifest bytes), so the embedded `signature`
field in the manifest was removed. **Back up the private key** — losing it means shipping a new app to rotate.
### 8.1 Fork created (2026-09-30)
* Fork: `Corfirean/azerothcore-wotlk-coa` (from `jealous-sound/azerothcore-wotlk-coa`). Default branch is `coa-bots`
(a branch named `coa` is impossible: the fork inherits upstream's `coa/...` branches). `main` stays a pure mirror of upstream.
* `coa-bots` = upstream `main` + a small, reviewable patch series (audit of the owner's 38 local commits, 2026-09-30):
1. `Core/Bots` hooks: `LFGMgr::GetProposalIdForPlayer`, `PlayerScript::OnPetitionOffered`, `Guild` friend `BotMgr`.
2. Crash fixes seen under bot load: `CombatManager::PutReference`, `Unit::RemoveFromWorld` dangling delayed-visibility pointer,
`InstanceMap::UnloadAll` teleport race, `IsStatisticAchievement` missing category.
3. `AscensionClassServiceBridge` + `AscensionResourceQuery` (needed by the bot module) and the socketless-bot appearance-sync guard.
4. RA: no throw on abrupt client disconnect (the Manager's RA client does exactly that).
5. CI workflow `coa-bots-build.yml`: builds core + bots module on Windows, runs `worldserver --version`.
* Discovery: the bot chassis (`LoginQueryHolder`, `IWorld::AddQueryHolderCallback`, `Group::GetRolls`) is already in upstream
(`feat(Core): add mod-playerbots hooks and CoA API (#3069)`), so it is not carried; `docs/core-patches.md` in the bots repo is outdated.
* Deliberately NOT carried (owner's other local work, decide separately): CoA mechanic corrections, `AscensionClassTester`
(dev harness; couples core to the bots module), respec `SpecId==0` guard, `cs_daynight`, aura/spell null-guards in
`SpellAuras.h`/`Spell.cpp`, class-contract data, debug `LOG_ERROR` traces in Petition/Query handlers, fleet log appenders in
`worldserver.conf.dist`, `apps/coa-dbc` binaries, `height_query` tool.
* Sync policy: `main` fast-forwards from upstream automatically; `coa-bots` is rebased/merged onto it nightly by a workflow in the
build repository (conflict => issue, no release). Channels: `edge` (every green nightly) and `stable` (promoted by the owner).
---------------------------------------------------------------------------------------------------------
## 9. Backlog (owner requests)
* **Interface languages: English (default), Russian, German, French, Spanish** (requested 2026-09-30).
Plan: i18n layer in the React UI (string catalogue per locale, ICU-style plurals, locale picked from the OS with a
switch in Settings, persisted per user); all user-facing strings currently written in English move into catalogues;
the Rust error catalogue (`error.rs` titles/messages/actions) is keyed by `ErrorCode` and translated on the UI side
(backend keeps stable codes; technical details stay untranslated for bug reports); dates, sizes and numbers use
`Intl`; settings-schema titles/descriptions (`schemas/*.json`) get per-locale overlays (`schemas/i18n/<locale>.json`);
translations are reviewed by native speakers before release, with an English fallback for missing keys and a CI check
that every locale has every key. Do this before the first public release so strings are not retro-fitted.
**Status (2026-09-30, updated):** catalogue + picker + every screen translated (en/ru/de/fr/es drafts); settings-schema titles,
descriptions, categories, presets, options and units are translated through overlays in `src/i18n/schema/`; diagnostics
checks, update steps, companion sizes and friends notes are recognised and translated in the UI. `tools/check-i18n.mjs`
verifies keys, placeholders and overlay coverage. Still English: free-form backend technical messages. Native-speaker
review is still needed (see `docs/TRANSLATIONS.md`).
* **Language choice on the very first screen** (requested 2026-09-30): the welcome screen must offer the language picker
right away (before any other text is read), preselected from the OS language; the choice is persisted and shared with
Settings. **Done 2026-09-30:** shared `LanguagePicker` (top-right of the welcome screen and in Settings).
* **Clean-install gap found by the owner (2026-09-30):** installing from the default URL fails with 404 because the `base`
release does not exist yet (only `edge` is published). The message is now human ("package not available yet"). Also open:
the base package has no `Data\` (maps/dbc/vmaps, ~3.8 GB); the installer needs a "use my existing Data folder / download"
step, and the owner must decide hosting/legal for game data before `base` is published.
* **Modules tab** (requested 2026-09-30): install/manage optional server modules (bots, AH bot, Solocraft, ...).
Design decision: AzerothCore modules are compiled into `worldserver.exe`, so "install" cannot mean "drop files in".
Plan: CI builds one binary containing a curated set of modules, every one disabled by default; the Manager only
enables/disables and configures them (`Core/configs/modules/<name>.conf`, schema-driven settings editor as for bots).
A `modules.json` catalogue travels with update packages (id, title, description, category, source repo + commit,
licence, conf file, enable key, settings schema, needed DB migrations, conflicts, status tested/experimental).
Adding a module = add it to the CI build list + catalogue entry, released as an ordinary update. Only modules tested
against the CoA fork and licence-compatible (AGPL/GPL, because the binary is distributed) are listed. The fork's own
CoA features (dynamic XP, ethereal bazaar, ...) appear in the same tab as "built in" toggles.
* **Known launcher leftovers:** the repack's `Scripts/manage.py` still renders `mod_ascension_compat.conf` at every start;
the core now ignores it ("is no longer read ... move its settings into coa.conf") - harmless warning, clean up together
with a launcher refresh. Migration-runner root cause for the re-created `item_template_ascension_compat` table (a module
script older than the rename is seen as pending) is worked around in `clean-base`, not fixed in the runner.
* **Bug found by the owner on the first clean install (2026-09-30):** login worked at the auth step but the game dropped
the connection. Causes: (1) the package shipped the maintainer's Trace-level logging templates (gigabytes of log per
hour, world thread stalled); (2) no active `coa.conf` (`CoA.Enable = 1`) because packages contain only `*.conf.dist`.
Fixed: `clean-base` restores quiet templates, `assemble-tree` ships `coa.conf.dist`, and the Manager creates missing
module `.conf` files from their `.dist` on install and before each start of a Manager-installed server.
* **Game client download and update through the Manager** (built 2026-10-01, `coa-core::clientdl`). The community PTR launcher
("Conquest of AzerothCore" Electron app) reads its client from
`https://launcher-api.coa-development.org/downloads/client/latest.json` (`{schema, version, publishedAt, files:[{path,size,sha256}]}`;
~200 files, 43 GiB: `Ascension.exe`, DLLs and ~40 big `Data/*.MPQ`) and fetches each file as the content-addressed object
`.../downloads/client/objects/<sha256>`. The Manager uses the same endpoints with its own resumable, hash-verified downloader:
* **State** lives in the client folder (`.coa-manager/client-state.json`: manifest version, per file size/mtime/sha256). A
routine check (every 5 minutes, like the server) is one small manifest request compared with the recorded version; no game file is read.
* **Plan, then apply** (like server updates): files are compared by size and mtime against the state, and hashed only when those
no longer match. Each file is `Missing`, `Changed` (still as the Manager left it) or `Modified` (the player changed it, e.g. a
custom `d3d9.dll` or a patched exe). Modified files are kept on request (and remembered: asked again only when the published file
changes) or replaced after being moved to `.coa-manager/replaced/<time>/`. Objects are staged in `.coa-manager/staging/<sha>` and
moved into place only after the hash matches; state is saved after every file, so cancel, crash or reboot resume where they stopped.
* **Never touched:** `WTF`, `Cache`, `Errors`, `Logs`, `*.log`, `realmlist.wtf`, `Interface/AddOns/CoABotUI` (those have their own paths).
* **UI:** PLAY becomes **UPDATE CLIENT** when a tracked client is outdated (with "Play without updating"); with no client linked it
becomes **Set up game client** (choose an existing folder, or download into `<chosen folder>/CoA Client` with a free-space check).
An existing client that is not tracked can be adopted from Settings > Game client ("Keep this client up to date").
A fresh download gets `realmlist.wtf` for each installed locale (the manifest has none) and, for local hosts, the companion addon.
* **Joining another server:** the welcome screen offers hosting or joining. Joining uses a separate player profile in
the Manager's `remote-client/connection.json`, storing the host address and client folder without a server registry entry,
installation metadata, database or server processes. Only Overview and application Settings appear in the sidebar.
Downloading, adopting and updating clients use the same verified downloader; Play applies the saved realmlist without
forcing a local realm selection. The player can return to hosting without removing their saved connection.
* **Open points:** ask the launcher maintainers before putting heavy load on their API; DBCs live inside the MPQ patch archives
(extracting them would need an MPQ reader), so the server's `Data\dbc` still comes from our `base` package for now.
+26
View File
@@ -0,0 +1,26 @@
# Host a CoA server for other players
**Portable characters are experimental.** The feature is in active development and may contain bugs.
## Set up your server
1. Choose **Host a server** on the welcome screen.
2. Choose **Install new server**, or **I already have a server**.
3. Finish setup, then start the server from **Overview**.
4. Link your game client and test joining your server yourself.
## Play with friends
Open **Play with friends**. Use the LAN/private-network instructions for your connection method and run the connectivity checks. Your computer and server must stay on while friends play. Share the connection information prepared by the Manager, never database credentials.
## Let other players find your realm
1. Open **Settings**, then find **Public listing**.
2. Fill in a clear realm name, a short description, language and region.
3. Review the join/account options and connection route.
4. Publish when you are ready to accept players.
5. Check that the realm appears in the browser and ask a friend to test joining from another connection.
A listing makes your server discoverable; it does not move hosting to the cloud or guarantee that your network can be reached. If joining fails, use the connection checks before changing router settings. Portable characters also require a compatible server build and its required settings.
Use **Backups** before changing server versions or optional modules. To stop accepting visitors, turn off the listing; to stop hosting, stop the server.
+43
View File
@@ -0,0 +1,43 @@
# Take your character with you
**Portable characters are experimental.** The feature is in active development and may contain bugs.
Portable characters let you keep a saved character in the Manager and play it on compatible CoA realms.
## Make your first character portable
1. Log out of the character in the game. Keep the Manager open.
2. Open **Characters** and choose **Make an existing character portable**.
3. Select a realm you can access and choose your character.
4. Choose **Make portable**. Your character appears under **Your characters**.
If a character cannot be selected, read the reason beside it. Bot characters, active challenges and some pending rewards can block the operation. Creating the portable copy does not delete the original realm character.
## Make a character from another realm portable
1. Open **Servers** and select the realm where your character lives.
2. Choose **Link an existing account** and enter the account you already use on that realm.
3. Log in to that realm in the game and stay at the character selection screen.
4. Switch back to the Manager with **Alt+Tab** and choose **Show my characters**.
5. Find your character and choose **Make portable**.
6. Wait for the confirmation and the **Portable** badge.
![Link your realm account](screenshots/09-link-realm-account.jpg)
![Choose your character](screenshots/10-remote-portable-candidates.jpg)
![Character saved](screenshots/11-remote-portable-saved.jpg)
## Play your portable character
1. Find a compatible server under **Servers** in player mode, or **Play with Friends** when hosting. You can also use a realm already available to your Manager.
2. Choose your portable character and review compatibility before pressing **Play**.
3. Follow the account prompt. Some hosts create an account for you; others require an existing account.
4. Launch the game when the Manager says the character is ready.
5. Keep the Manager open while playing. Log out and wait for progress to finish saving before closing it or moving to another realm.
A realm with a lower level cap may play your character at a lower level. Check the warning: unsupported content is held back rather than silently discarded. Portable play requires a compatible server; installing the Manager alone does not add support to every server.
## Saved versions and conflicts
Use **Saved versions** to see when progress was saved. If the Manager reports a conflict, follow its explanation before continuing. Avoid playing two copies at once, and keep normal server backups as well.
+12
View File
@@ -0,0 +1,12 @@
# Guides and screenshots
- [Screenshot gallery](SCREENSHOTS.md) — explore the interface and its main features.
- [Host your own server](HOST-A-SERVER.md) — install a server, invite friends and publish a realm listing.
- [Portable characters · Experimental](PORTABLE-CHARACTERS.md) — save an existing character and use compatible realms.
- [Wildcard](WILDCARD.md) — switch realm profiles and manage separate characters.
- [Warcraft Reborn](WARCRAFT-REBORN.md) — create and switch to a Warcraft Reborn realm.
- [Updates, backups and repair](REPAIR_AND_REALMS.md) — keep your server updated and recover from problems.
**Portable characters are experimental.** The Manager is in active development and bugs may still occur. Please [report problems](https://github.com/Corfirean/coa-server-manager/issues/new/choose).
[Download the Manager](https://github.com/Corfirean/coa-server-manager/releases/latest) · [Manager changelog](../CHANGELOG.md) · [Server changelog](https://github.com/Corfirean/coa-server-build/blob/main/CHANGELOG.md)
+17 -30
View File
@@ -1,44 +1,31 @@
# Database checks, repair and simultaneous worlds
# Updates, backups and repair
Settings includes **Check files and database**, **Repair**, and **Start CoA and Wildcard together**.
## Update your server
Database checks inspect both realm schemas when Wildcard exists, the migration ledger, character-save columns and CoA starting class data. Results are saved in the installation metadata's `logs/database-checks.json` and included in exported diagnostics. Hash checks skip mutable MySQL data and configuration files.
Select your server in the Manager. When an update is available, choose **UPDATE** and follow the prompts. Keep the Manager open until it finishes: it stops the server, creates a recovery point and checks the updated server.
Repair requires both worlds and auth to be stopped, no unfinished update, and the installed version to match the currently published signed update. The signed base must match the installed base as well. It verifies packages before changing files, creates a full database/config backup, saves every replaced file and a recovery inventory under `<server>.manager/repairs/<backup-id>/`, then restores official files and applies pending migrations to both worlds. Settings and player databases are not replaced. A failure during file replacement restores the original files. Interrupted repair can be retried; its file inventory and backups remain available.
Updates preserve accounts, characters and settings and cover every realm you have created. If an update fails, read the displayed error before trying again; recovery information remains available in the Manager.
Repair reports unresolved schema damage. It does not guess ALTER statements or replay already applied SQL: those cases need a new corrective migration. Applied SQL whose recorded SHA-256 changed is reported and rejected. Manager failure records take precedence over the core's old `updates` entries.
## Keep a backup
## Release schema contracts
Use **Backups** to save characters, accounts and settings before important changes. Full backups also include world data. Check the realm recorded in a recovery point before restoring it, and stop the server before a restore.
`clean-base` clears the character migration ledger after rebuilding the character database and captures `Scripts/database-schema.json`. It contains expected table/column names and column types. Installation and migration validation inspect this contract; older packages receive the narrower character-save checks, clearly labelled in the UI.
Portable character saved versions do not replace server backups. **Portable characters are experimental.**
For each update, capture a fresh contract **from a disposable database already migrated to the exact packaged core**, into the release tree before `pack-update`:
## Check or repair an installation
```powershell
cargo run -p coa-release -- extract-schema-base --package C:\games\signed-base --fixture C:\games\coa-schema-fixture-build
cargo run -p coa-release -- schema-contract --repack C:\games\coa-schema-fixture-build --tree C:\games\release-tree --core C:\games\release-core --bots C:\games\release-bots --repairs C:\games\release-repairs
```
1. Stop the server.
2. Open **Settings** and run **Check files and database**.
3. Review the results. If official server files need restoring, use **Repair** and follow the prompts.
The first command verifies every entry in the signed base archive and extracts only the runtime needed for a disposable database. The second applies the release SQL with temporary ports, checks critical feature tables independently of migration history, and captures the resulting schema. It requires the disposable fixture marker and never starts a world server.
Repair creates a recovery backup and restores official files while preserving settings and player databases. It may require the installed version to match the currently published signed package. Missing player data requires a backup; some database problems require a corrective server update.
Base and update packaging require a complete contract for all three databases. Cumulative updates always carry it, even when unchanged. Signing and verification check the archive contents and contract before publication. Never capture it from a user's server or from an unmigrated previous release database.
If an operation stays blocked or fails again, use **Report a problem** and include diagnostics and the error message. Do not delete database files or recovery records to clear an error.
Corrective SQL under `repairs/{auth,characters,world}` receives new `manager_repair__` migration IDs and runs after historical SQL. Missing Wildcard tables can therefore be restored without replaying historical data deletion. Existing malformed tables still fail schema checks; creating a missing table does not recover its lost rows.
## Run CoA and Wildcard together
## Simultaneous worlds
In **Settings**, enable **Start CoA and Wildcard together**. Both realms share accounts but keep separate characters and world data. Both statuses appear on Overview, and Stop shuts down both.
On Windows the checkbox creates Wildcard when needed and saves distinct secondary world/RA ports. Both worlds start with one shared authserver/MySQL and separate world/character databases. The second world has its own executable copy, config, supervisor, logs and report spool in `.realms/secondary`; Data is read through the primary installation's absolute path. This runtime folder is excluded from packages.
If friends networking was configured before enabling the second realm, run its setup again so both realms are reachable. Disable simultaneous startup before switching to Warcraft Reborn.
The selected realm remains the primary world used by the Manager's console, companions, settings and Play. Selecting another realm still stops/restarts both to preserve that relationship. CoA companion restrictions on Wildcard remain in force. Both world statuses are shown on Overview.
Stop shuts down the secondary supervisor/world before the primary launcher may stop the shared database. Database restores, updates and profile changes reject a running secondary world. Each start refreshes its executable/config copies, so server updates cannot leave an old secondary binary behind. The secondary world inherits the primary network bind setting; its RA console remains on loopback. Enabling friends networking opens/maps both world ports. If networking was enabled before the second realm, enable it again to add the second port's rule/mapping.
## Disposable probes
```powershell
cargo run -p coa-core --example multiworld_e2e -- C:\games\your-fixture
cargo run -p coa-core --example repair_e2e -- C:\games\your-fixture
cargo run -p coa-core --example schema_e2e -- C:\games\your-fixture
```
The first checks both processes, ports, shared services and graceful shutdown. The second uses a test-only signing key, corrupts a fixture file, applies SQL to both worlds, verifies backups and account/character counts, and drops a column only in Wildcard to prove schema checks are isolated. These probes modify the fixture; never use a real server.
[All guides](README.md)
+41
View File
@@ -0,0 +1,41 @@
# CoA Server Manager 0.7.0
### Your server, a few clicks away
![Your server, a few clicks away](screenshots/setup.png)
### Your server at a glance
![Your server at a glance](screenshots/overview.png)
### Build your own adventuring company
![Build your own adventuring company](screenshots/bots.png)
### Manage your server modules
![Manage your server modules](screenshots/modules.png)
### Your client, ready to play
![Your client, ready to play](screenshots/client.png)
### Keep your progress safe
![Keep your progress safe](screenshots/backups.png)
### Bring your friends along
![Bring your friends along](screenshots/friends.png)
### Find your next realm
![Find your next realm](screenshots/realms.png)
### Portable characters · Experimental
![Portable characters · Experimental](screenshots/portable.png)
### Make a character portable · Experimental
![Make a character portable · Experimental](screenshots/remote-character.png)
-36
View File
@@ -1,36 +0,0 @@
# Translations
The Manager interface exists in English (source), Russian, German, French and Spanish. Everything except English is a
**draft written by an AI assistant and needs review by a native speaker** before it is treated as final.
## Where the texts are
| What | File | Notes |
|---|---|---|
| Interface texts | `src/i18n/locales/<lang>.ts` | flat `"key": "text"` lists; `en.ts` is the source of truth, other languages fall back to it |
| Setting titles, descriptions, categories, presets, units, option labels | `src/i18n/schema/<lang>.json` | generated once from `schemas/*.json`; edit the JSON directly |
| Backend error messages | keys `err.<code>.title` / `err.<code>.message` in the locale files | codes come from `crates/coa-core/src/error.rs` |
`{name}` placeholders must stay exactly as they are. Plural forms use the suffixes `_one`, `_few`, `_many`, `_other`
(the browser picks the right one for the language).
## Checking your work
```
node tools/check-i18n.mjs # unknown keys, changed placeholders, coverage, schema overlays complete
npx tsc --noEmit # types
npx vite # UI with mock data in a browser; pick the language on the welcome screen
```
## Adding a language
1. Copy `src/i18n/locales/en.ts` to `<lang>.ts` (keep the keys), translate, and register it in `src/i18n/index.tsx`
(`Locale`, `LOCALES`, `DICTS`).
2. Copy `src/i18n/schema/ru.json` to `<lang>.json`, translate the values, and register it in `SCHEMA`.
3. Run the check above.
## Known gaps
* Free-form technical messages produced deep inside the backend (`Error::Invalid("...")`) stay in English; they are
shown as "technical details".
* Texts inside the game itself (client, server data) are not part of the Manager.
+16
View File
@@ -0,0 +1,16 @@
# Warcraft Reborn
Choose **Warcraft Reborn** on Overview to create or switch to this realm. The
installed worldserver must contain Warcraft Reborn support; an old build is
refused before any databases are created. Stop the client before switching to
have its saved realm updated immediately.
Accounts are shared with your other local realms. Warcraft Reborn starts with separate characters; switching profiles does not copy existing characters or progress.
Bots are unavailable on Warcraft Reborn. Other optional modules are experimental. Portable character transfers into Warcraft Reborn are not supported. **Portable characters remain experimental.**
Backups and updates include every created realm. Check the realm recorded in a recovery point before restoring it.
Simultaneous startup supports CoA + Wildcard only. Disable **Start CoA and Wildcard together** in Settings before switching to Warcraft Reborn.
[All guides](README.md)
+1 -17
View File
@@ -23,20 +23,4 @@ adapter before the next start.
![Module compatibility](screenshots/wildcard-demo/modules.jpg)
## Validation for v0.5.0
- `cargo test --workspace`: 178 core tests passed; command layer and examples compiled.
- `npm run build`, translation key checks and changelog validation passed.
- Native disposable fixture on offset ports: fresh Wildcard creation and startup, CoA startup after switching back,
separate probe values and restored CoA configuration. A second fresh creation verified the arena-season seed fix.
- Full backup verified both worlds, both character databases, shared auth and configurations.
- Native migration probe applied once to each realm and once to auth; repeating it did nothing.
- Restoring Wildcard characters removed a fixture-only change and preserved the CoA probe value.
- Core gameplay wrapper: `VERIFY ALL: PASSED`, gameplay stage only, real clock. First login (7 assertions),
level rolls/reroll (31), skill card rolls (13), stat path (12). The level-roll and skill-card scenarios passed on isolated
reruns after failing in the shared batch; the wrapper records two batch-sensitive cases.
- Browser demo verified active profile, hidden Bots navigation and module compatibility labels. Screenshots show
mock IPC; native checks above exercised the actual backend and server.
The gameplay harness required fixture-only synchronization of already applied migration history by verified SQL
hashes and its normal Windows module config path. Owned server files and databases were not modified.
[All guides](README.md)
Binary file not shown.

After

Width:  |  Height:  |  Size: 62 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 67 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 75 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 68 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 71 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 106 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 82 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 74 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 88 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 90 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 91 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 58 KiB

After

Width:  |  Height:  |  Size: 357 KiB

Loaded 100 of 127 files, more files were not shown because too many files have changed in this diff. Show more