docs(logbook): record work items incident audit production verification [skip ci]
This commit is contained in:
@@ -53,7 +53,7 @@ local browser smoke:
|
||||
- Kali `192.168.0.112` 仍沒有執行 full-upgrade、autoremove、reboot、active scan 或 `/execute`。
|
||||
- 下一個真正推動 headline progress 的 Gate 仍是負責人回覆被驗收、redacted finding ingestion、GitHub primary evidence 或 active runtime gate 其中之一。
|
||||
|
||||
## 2026-05-31|Work Items 焦點 Incident 稽核鏈 local 收斂
|
||||
## 2026-05-31|Work Items 焦點 Incident 稽核鏈 production 收斂
|
||||
|
||||
**背景**:
|
||||
|
||||
@@ -92,18 +92,71 @@ http://127.0.0.1:3107/zh-TW/awooop/work-items?project_id=awoooi&incident_id=INC-
|
||||
horizontalOverflow=false
|
||||
```
|
||||
|
||||
**Gitea / Production deploy**:
|
||||
|
||||
```text
|
||||
e6a433da fix(web): surface incident audit chain in work items
|
||||
ab780892 chore(cd): deploy 7987da7 [skip ci]
|
||||
59b4943b feat(web): 視覺化 IwoooS 資安指揮板
|
||||
af70ce8e chore(cd): deploy 59b4943 [skip ci]
|
||||
|
||||
k8s awoooi-prod:
|
||||
awoooi-web 192.168.0.110:5000/awoooi/web:59b4943bf99be8be1ae8be3f2f640467964744a7 2/2
|
||||
awoooi-api 192.168.0.110:5000/awoooi/api:59b4943bf99be8be1ae8be3f2f640467964744a7 2/2
|
||||
awoooi-worker 192.168.0.110:5000/awoooi/api:59b4943bf99be8be1ae8be3f2f640467964744a7 1/1
|
||||
|
||||
CI/CD:
|
||||
59b4943b build-and-deploy=success
|
||||
59b4943b post-deploy=success
|
||||
summary=API=✅; Web=✅; AlertChain=✅; SourceLink=✅; Monitoring=✅; Smoke=⚠️
|
||||
```
|
||||
|
||||
**Production 驗證**:
|
||||
|
||||
```text
|
||||
https://awoooi.wooo.work/zh-TW/awooop/work-items?project_id=awoooi&incident_id=INC-20260530-0DD83C
|
||||
visible: 焦點事件稽核鏈
|
||||
visible: 處理流程
|
||||
visible: 執行與學習證據
|
||||
visible: webhook:ok > investigator:ok > ai_router:ok > llm:ok > target:ok > blast:ok > safe:ok > executor:fail > verifier:warn > km:ok
|
||||
visible: Automation: ansible_check_mode_executed
|
||||
visible: infra/ansible/playbooks/188-ai-web.yml
|
||||
visible: KM entry written
|
||||
visible: incident_open_after_successful_execution
|
||||
canScroll=true
|
||||
horizontalOverflow=false
|
||||
|
||||
production API:
|
||||
/api/v1/platform/status-chain?incident_id=INC-20260530-0DD83C:
|
||||
current_stage=execution_succeeded
|
||||
verdict=auto_repaired_verification_degraded
|
||||
verification=degraded
|
||||
latest_operation_type=ansible_check_mode_executed
|
||||
latest_returncode=2
|
||||
latest_apply_executed=false
|
||||
/api/v1/incidents/INC-20260530-0DD83C/timeline:
|
||||
stage_total=11
|
||||
event_total=18
|
||||
```
|
||||
|
||||
**Post-deploy 風險記錄**:
|
||||
|
||||
- 59b rollout 後 web pods 各有 `restartCount=2`;K8s events 顯示 startup / readiness / liveness probe 曾 timeout,期間瀏覽器一度命中 `502 Bad Gateway`。
|
||||
- 重新等待 rollout settled 後,`kubectl rollout status deploy/awoooi-web` 成功,`/api/v1/health` 200 healthy,正式站 Work Items 重新驗證通過。
|
||||
- 這是 Smoke=⚠️ 的合理風險來源,後續應追 web startup/liveness probe budget 或 Next cold-start profile;本輪沒有用手動 K8s hotfix 掩蓋問題。
|
||||
|
||||
**技術債 / 現場清理**:
|
||||
|
||||
- 本機 `/System/Volumes/Data` 一度只剩約 103MiB,導致 Git 無法寫入 `FETCH_HEAD`,且 build 先前出現 webpack cache ENOSPC 警告。
|
||||
- 已只清理本 worktree 產生物 `apps/web/.next`,釋放約 1.5GiB;未刪資料庫、原始碼或任何 production 狀態。
|
||||
- 後續仍應把 local runner / build cache 空間列為開發機維運項,避免前端驗證被快取空間污染。
|
||||
|
||||
**目前整體進度(local ready, pending Gitea deploy)**:
|
||||
**目前整體進度(post-deploy)**:
|
||||
|
||||
- Telegram / Run / Work Items 單一 Incident drill-down:約 88%;Run detail 已 production,Work Items 已 local 驗證,待推版與 production smoke。
|
||||
- MCP / Sentry / SigNoz / KM / PlayBook / Ansible 的跨頁透明度:約 84%;Work Items 已能承接同一條 timeline,但 Approvals / Tickets 還需同樣接入。
|
||||
- 前端 AI 自動化管理介面同步:約 85%;工作鏈路頁開始成為操作員入口,不再只靠 Telegram 按鈕。
|
||||
- 整體 AI 自動化飛輪:約 73%;仍不能宣稱 24h 全自動 repair 閉環,需以 production evidence 持續補齊。
|
||||
- Telegram / Run / Work Items 單一 Incident drill-down:約 94%;Run detail 與 Work Items 都已 production 驗證同一 Incident timeline。
|
||||
- MCP / Sentry / SigNoz / KM / PlayBook / Ansible 的跨頁透明度:約 88%;Work Items 已承接同一條 timeline,但 Approvals / Tickets 還需同樣接入。
|
||||
- 前端 AI 自動化管理介面同步:約 89%;工作鏈路頁開始成為操作員入口,不再只靠 Telegram 按鈕。
|
||||
- 整體 AI 自動化飛輪:約 74%;仍不能宣稱 24h 全自動 repair 閉環,需以 production evidence 持續補齊。
|
||||
- 24h 完整 AI Agent 自動修復 production claim:0%;仍維持嚴格口徑,只能宣稱「已驗證的特定 controlled apply / drill-down 能被追蹤」。
|
||||
|
||||
## 2026-05-31|Ollama health 診斷碼與前台冷卻顯示
|
||||
|
||||
Reference in New Issue
Block a user