fix(k8s): cap api replicas until db role limit is raised [skip ci]

This commit is contained in:
Your Name
2026-07-01 16:28:19 +08:00
parent 6f1feef6a2
commit 48d2d0f498
2 changed files with 8 additions and 2 deletions

View File

@@ -31,17 +31,19 @@
- ArgoCD refresh 後成功同步到 `9db4f72cb`,但 API rollout 卡在 1/2Pending pod event 顯示 `nodeSelector=mon` 搭配 `topologySpreadConstraints minDomains=2 whenUnsatisfiable=DoNotSchedule` 自相矛盾,造成 scheduler deadlock。
- 已把 API / Web / Worker 的 topology spread 從硬性 `DoNotSchedule + minDomains=2` 調整為柔性 `ScheduleAnyway`,保留 hostname spread preference但不再阻塞 post-reboot / GitOps rollout。
- API 新 pod 排程後又 `CrashLoopBackOff`previous log 顯示 `harbor_registry_controlled_recovery_preflight.py` 在 production container layout `/app/src/services/...` 使用 `Path(__file__).parents[4]` 造成 `IndexError: 4`。已改用既有 `resolve_repo_root(Path(__file__))`,並同步修掉同型的 StockPlatform controlled recovery preflight 路徑解析,避免下一個 import 再炸。
- 最新 API image `54ac3cd...` 啟動後又因 `asyncpg.exceptions.TooManyConnectionsError: too many connections for role "awoooi"` CrashLoopreadback 顯示 Postgres `max_connections=300`,但 `awoooi` role `rolconnlimit=2`,且 `awoooi_migrator``ALTER ROLE` 權限。非 secret admin path 目前不可用,因此先把 API `replicas` 暫時降為 `1`,讓最新版 API 在既有 role limit 內恢復;`replicas=2` 必須等 DB role limit / pool budget 修復後再恢復。
**驗證**
- `argocd-repo-server` rollout`deployment "argocd-repo-server" successfully rolled out`
- ArgoCD readback`sync=Synced rev=9db4f72cbf98`API deployment image / env 已更新為 `06819ea96c05...`
- `kubectl apply --dry-run=server` 驗證 `awoooi-api``awoooi-web``awoooi-worker` manifests通過。
- `py_compile` / targeted pytest 需在 path resolver fix 後重跑並推正常 CD因為 production 必須建出新 API image 才能離開 CrashLoop。
- DB role readback`awoooi rolconnlimit=2``awoooi_migrator rolconnlimit=-1` 但無 `ALTER ROLE` 權限188 local postgres admin path 需要密碼,未讀 secret。
**邊界**:未讀 secret / token / `.env` / raw sessions / SQLite / auth未使用 GitHub / `gh` / GitHub API未重啟主機 / Docker / Nginx / K3s / DB / firewalllive patch 僅限 ArgoCD repo-server init command 與 prod Deployment topology scheduler policy。
**下一步**
- 推 Gitea main 後等 CD 建新 API image,讀回 API / Web / Worker 全部 Ready、production workbench runtime/desired tag 等於最新 commit再判定 Gitea/CD 恢復完成
- 推 Gitea main 後等 ArgoCD 套用 `replicas=1` recovery cap,讀回 API / Web / Worker Ready、production workbench runtime/desired tag 等於最新 deploy marker後續 P0 必須修 DB role limit / pool budget 後恢復 API `replicas=2`
## 2026-07-01 — 15:25 Gitea live recovery after 110 Docker/control-plane failure