docs(recovery): record non110 runner lane recovery [skip ci]

This commit is contained in:
Your Name
2026-07-01 08:20:52 +08:00
parent 693a31343b
commit 52d91d256d

View File

@@ -1,3 +1,23 @@
## 2026-07-01 — 08:19 188 non-110 CD lane controlled apply / 110 Harbor blocker readback
**照主線修正的問題**
- 最新 Gitea main fast-forward 到 `693a31343 feat(agent): expose harbor receipt output contract`,本地 focused 驗證通過後讀回 live queue。
- Gitea queue 一度從 110 `awoooi-host` blocker 變成 `awoooi-non110-host` / `awoooi-non110-ubuntu` no-matching表示非 110 CD lane 本身也離線;這是同一條 deploy 主線,不是新支線。
- 188 readiness verifier 先回 `READY_CONFIG_COUNT=1``READY_BINARY_COUNT=1``READY_REGISTRATION_COUNT=1``READY_SERVICE_COUNT=1`,但 `READY_ACTIVE_SERVICE_COUNT=0``AWOOOI_NON110_RUNNER_READY=0`registration metadata 存在且 `content_read=false`service target binary / config verifiedhost pressure OK。
- 執行受控 apply`AWOOOI_NON110_ENABLE=1 bash -s -- --enable < ops/runner/install-awoooi-non110-runner-user-service.sh` on `ollama@192.168.0.188`。讀回 `SERVICE_ENABLE_PERFORMED=1``SERVICE_START_PERFORMED=1``awoooi-non110-runner.service ActiveState=active``MainPID=3224103`、rollback unit / autostart path present未讀 token 或 `.runner` 內容。
- apply 後 rerun readiness`AWOOOI_NON110_RUNNER_READY=1``BLOCKER_COUNT=0``READY_ACTIVE_SERVICE_COUNT=1``READY_AUTOSTART_PATH_COUNT=1`、load/core `0.901667`
- `verify-awoooi-non110-cd-closure.py --readiness-file /tmp/awoooi-non110-readiness-20260701-081558.txt --json``non110_runner_ready=true`,整體 closure 正確停在 `blocked_harbor_110_repair_no_matching_runner`
- 最新 CD `#4191` 已能在 non-110 lane 跑到 Harbor login retryfinal `Failure`classifier=`harbor_registry_public_route_unavailable`、registry `/v2` status `502`Harbor repair `#4190``Waiting`no-matching label `awoooi-host`
- Read-only cold-start artifact `/tmp/awoooi-cold-start-mainline-20260701-080527.log``PASS=64 WARN=7 BLOCKED=5`,第一順位 blockers 仍是 110 registry `/v2`、110 SSH read-only check、K3s registry pull refused by `110:5000`
- 110 read-only metrics 顯示 `awoooi_host_gitea_actions_active_container_count=0`、runner fail-closed enforcer active、`root_restore_sources_left=0`host load 仍 high`gitea` container 約 `3.4019` cores。下一步不是恢復 legacy / generic runner而是 110 local console / root shell 跑 `recover-110-control-path-and-harbor-local.sh --check` / controlled lane verifier再按 receipt 補 Harbor watchdog verifier。
**驗證**
- `DATABASE_URL=postgresql+asyncpg://test:test@localhost:5432/test PYTHONPATH=apps/api python3.11 -m pytest apps/api/tests/test_harbor_registry_controlled_recovery_receipt.py apps/api/tests/test_awoooi_priority_work_order_readback_api.py apps/api/tests/test_ai_agent_log_controlled_writeback_executor_readback_api.py apps/api/tests/test_ai_agent_autonomous_runtime_control.py ops/runner/test_read_public_gitea_actions_queue.py scripts/reboot-recovery/tests/test_reboot_auto_recovery_slo_scorecard.py scripts/reboot-recovery/tests/test_cold_start_monitor_bounded_probes.py -q` 通過79 passed
- `ruff check``py_compile``ops/runner/guard-gitea-runner-pressure.py --root .``scripts/ci/check-gitea-step-env-secrets.js``git diff --check` 通過。
- `python3.11 -m pytest ops/runner/test_check_awoooi_non110_runner_readiness.py ops/runner/test_register_awoooi_non110_runner.py ops/runner/test_guard_gitea_runner_pressure.py -q` 通過14 passed
**邊界**:只對 188 user-level non-110 runner service 執行 allowlisted controlled apply未讀 secret / token / `.env` / raw sessions / SQLite / auth`.runner` 只確認 metadata 存在且 `content_read=false`;未使用 GitHub / `gh` / GitHub API未 workflow_dispatch未重啟主機、未 restart Docker / Nginx / K3s / DB / firewall未使用 non-110 runner 修 110 Harbor。
## 2026-07-01 — 110 SSH high-load receipt blocker
**照主線修正的問題**