Compare commits
12 Commits
codex/110-
...
codex/depl
| Author | SHA1 | Date | |
|---|---|---|---|
|
5e37777c87 | ||
|
|
aaf3f7bfab | ||
|
|
0cb66f371c | ||
|
|
940adca8d5 | ||
|
|
ebb77719e2 | ||
|
|
8251026c06 | ||
|
|
5647e3e74c | ||
|
|
d4a676b7db | ||
|
|
b6c600e24d | ||
|
|
aa41db6875 | ||
|
|
7c3b1c0ab9 | ||
|
|
77ba5ed517 |
@@ -205,7 +205,8 @@ jobs:
|
||||
docker run --rm \
|
||||
--name "awoooi-cd-${GITHUB_RUN_ID:-manual}-${GITHUB_RUN_ATTEMPT:-1}-api-tests" \
|
||||
--cpus "2.0" \
|
||||
--memory "2g" \
|
||||
--memory "6g" \
|
||||
--memory-swap "8g" \
|
||||
-v "$PWD:/workspace" \
|
||||
-v /tmp/awoooi-api-tests.sh:/tmp/awoooi-api-tests.sh:ro \
|
||||
-v awoooi-api-venv-cache:/opt/api-venv \
|
||||
|
||||
@@ -1,3 +1,29 @@
|
||||
## 2026-06-28 — 02:06 110 runner fail-closed guard 轉 controlled automation
|
||||
|
||||
**背景**：統帥明確要求非 critical hard gate / guard 全部打開並快速推進正式部署。`2a1cd3cc8 fix(reboot): fail closed host runner startup` 將 110 startup runner path 改成 sentinel fail-closed，且 disabled 分支會 `disable --now` / `SIGKILL` / `pkill -KILL` 正在跑的 runner；live `/usr/local/bin/awoooi-startup-110.sh` 與 user-level runner service 也仍是舊 guard 版本，會重新阻斷 CD。
|
||||
|
||||
**完成內容**：
|
||||
- `scripts/reboot-recovery/awoooi-startup-110.sh` 將 AWOOI 專用 runner 預設改為 controlled automation 啟動：`AWOOOI_START_GITEA_RUNNER_ON_BOOT` 預設 `1`。
|
||||
- sentinel 改為 optional：只有 `AWOOOI_REQUIRE_RUNNER_ENABLE_SENTINEL=1` 時才要求 `AWOOOI_RUNNER_ENABLE_SENTINEL`。
|
||||
- disabled 分支不再預設殺掉既有 runner；只有明確設定 `AWOOOI_STOP_GITEA_RUNNER_WHEN_DISABLED=1` 才執行 stop / SIGKILL / pkill。
|
||||
- live 110 已同步 `/usr/local/bin/awoooi-startup-110.sh`、system `gitea-act-runner-host.service`、user-level `gitea-act-runner-host.service` 為受控版本；system runner service 已 `enabled` 且 `active/running`。
|
||||
- live 110 已移除 `/etc/systemd/system/gitea-act-runner-host.service.d/00-awoooi-disabled-pressure-guard.conf`、清掉 runner binary / service immutable bit，並保留原檔備份。
|
||||
|
||||
**本地 / live 驗證結果**：
|
||||
- `bash -n scripts/reboot-recovery/awoooi-startup-110.sh`：通過。
|
||||
- `git diff --check`：通過。
|
||||
- live `systemctl show gitea-act-runner-host.service`：`LoadState=loaded`、`ActiveState=active`、`SubState=running`、`Result=success`。
|
||||
- live `/usr/local/bin/awoooi-startup-110.sh`：讀回 `AWOOOI_START_GITEA_RUNNER_ON_BOOT="${AWOOOI_START_GITEA_RUNNER_ON_BOOT:-1}"`。
|
||||
- live user-level runner service：讀回 `ExecStart=/home/wooo/act-runner/act_runner daemon --config /home/wooo/act-runner/config.yaml`，不再是 `/bin/false`。
|
||||
|
||||
**仍保留的 break-glass 邊界**：
|
||||
- 本段只打開 AWOOI 專用 CD runner controlled automation；沒有放寬 secret value、private key、token、cookie、credential URL、raw `.env`、raw session / SQLite。
|
||||
- 沒有做 DB destructive / backup restore / force push / repo deletion / refs deletion / paid provider route switch / external active exploit scan。
|
||||
|
||||
**下一步**：
|
||||
- commit / push 本段 runner controlled automation patch。
|
||||
- 等最新 deploy marker 後讀回正式 Approvals、Runs、Work Items、Alerts，確認 AwoooP 低 / 中 / 高風險流程不再把人工當預設終局。
|
||||
|
||||
## 2026-06-28 — GitHub private backup controlled execution 授權 gate 本地完成
|
||||
|
||||
**背景**：統帥明確要求「硬閘全部打開、完全授權、全面快速推進」。本段把舊 GitHub private backup `blocked/read-only` gate 改成可審計的 owner controlled execution authorization；這是授權 gate 變更，不是秘密值收件，也不是已完成 GitHub 寫入。
|
||||
|
||||
@@ -41,7 +41,7 @@ resources:
|
||||
images:
|
||||
- name: 192.168.0.110:5000/library/api:IMAGE_TAG_PLACEHOLDER
|
||||
newName: 192.168.0.110:5000/awoooi/api
|
||||
newTag: 30af0fb4202073719f8c8011fe773f806576ede0
|
||||
newTag: 8251026c06f0707e7c955fce55dd8ddd8d5e7662
|
||||
- name: 192.168.0.110:5000/library/web:IMAGE_TAG_PLACEHOLDER
|
||||
newName: 192.168.0.110:5000/awoooi/web
|
||||
newTag: 30af0fb4202073719f8c8011fe773f806576ede0
|
||||
|
||||
@@ -7,11 +7,16 @@ set -euo pipefail
|
||||
# production host and a CI host, so CD must not start a new Docker/Next build
|
||||
# while load, BuildKit, Gitea Actions, or headless smoke pressure is already high.
|
||||
# This gate never kills, renices, or rewrites another repo's process tree.
|
||||
# 2026-06-28 Codex: CD trigger after opening the AWOOI runner pressure guard.
|
||||
# 2026-06-28 Codex: CD trigger after opening the AWOOI direct runner warn-only guard.
|
||||
# 2026-06-28 Codex: non-behavior trigger after restoring the quarantined runner binary.
|
||||
# 2026-06-28 Codex: non-behavior trigger after increasing API test container memory.
|
||||
# 2026-06-28 Codex: commander blanket authorization opens this guard to
|
||||
# short-wait warn-only by default; critical blockers stay outside this script.
|
||||
# 2026-06-28 Codex: retrigger after moving the runner to a direct transient unit.
|
||||
|
||||
ATTEMPTS="${HOST_WEB_BUILD_PRESSURE_ATTEMPTS:-${HOST_WEB_BUILD_PRESSURE_MAX_ATTEMPTS:-60}}"
|
||||
ATTEMPTS="${HOST_WEB_BUILD_PRESSURE_ATTEMPTS:-${HOST_WEB_BUILD_PRESSURE_MAX_ATTEMPTS:-6}}"
|
||||
SLEEP_SECONDS="${HOST_WEB_BUILD_PRESSURE_SLEEP_SECONDS:-${HOST_WEB_BUILD_PRESSURE_INTERVAL:-10}}"
|
||||
WARN_ONLY="${HOST_WEB_BUILD_PRESSURE_WARN_ONLY:-0}"
|
||||
WARN_ONLY="${HOST_WEB_BUILD_PRESSURE_WARN_ONLY:-1}"
|
||||
MAX_LOAD5_PER_CORE="${HOST_WEB_BUILD_PRESSURE_MAX_LOAD5_PER_CORE:-0.85}"
|
||||
MAX_CI_CPU_PERCENT="${HOST_WEB_BUILD_PRESSURE_MAX_CI_CPU_PERCENT:-250}"
|
||||
# One Gitea Actions task container/process group is the current job itself.
|
||||
|
||||
@@ -184,17 +184,25 @@ fi
|
||||
# ──────────────────────────────────────────────
|
||||
# STEP 6: Gitea Act Runner（CI/CD 核心）
|
||||
# 2026-04-05 Claude Code: 加入 — 解決重開機後 Gitea runner 離線、CD 失效
|
||||
# 2026-06-27 Codex: 110 是 production / registry / observability 主機；
|
||||
# runner 預設維持停用降壓，未完成限流 / 搬遷前不可在 startup 自動拉起。
|
||||
# 2026-06-27 Codex: 110 runner labels 收斂，避免接泛用 shared CI。
|
||||
# 2026-06-28 Codex: AWOOI runner labels 已收斂為專用 labels；
|
||||
# 非 critical CD runner gate 改為 controlled automation，避免 startup
|
||||
# script 誤殺正在執行的正式部署。sentinel 僅在明確要求時作為第二鑰匙。
|
||||
# ──────────────────────────────────────────────
|
||||
log "[6/6] 檢查 Gitea Act Runner（預設不自動啟動）..."
|
||||
log "[6/6] 檢查 Gitea Act Runner（預設受控啟動）..."
|
||||
RUNNER_DIR="/home/wooo/act-runner"
|
||||
RUNNER_SERVICE="gitea-act-runner-host.service"
|
||||
RUNNER_ENABLE_SENTINEL="/run/awoooi-runner-host-enabled"
|
||||
START_GITEA_RUNNER_ON_BOOT="${AWOOOI_START_GITEA_RUNNER_ON_BOOT:-0}"
|
||||
RUNNER_ENABLE_SENTINEL="${AWOOOI_RUNNER_ENABLE_SENTINEL:-/run/awoooi-runner-host-enabled}"
|
||||
START_GITEA_RUNNER_ON_BOOT="${AWOOOI_START_GITEA_RUNNER_ON_BOOT:-1}"
|
||||
REQUIRE_RUNNER_ENABLE_SENTINEL="${AWOOOI_REQUIRE_RUNNER_ENABLE_SENTINEL:-0}"
|
||||
STOP_GITEA_RUNNER_WHEN_DISABLED="${AWOOOI_STOP_GITEA_RUNNER_WHEN_DISABLED:-0}"
|
||||
START_GITEA_RUNNER_ALLOWED=0
|
||||
if [ "$START_GITEA_RUNNER_ON_BOOT" = "1" ] && [ -e "$RUNNER_ENABLE_SENTINEL" ]; then
|
||||
START_GITEA_RUNNER_ALLOWED=1
|
||||
if [ "$START_GITEA_RUNNER_ON_BOOT" = "1" ]; then
|
||||
if [ "$REQUIRE_RUNNER_ENABLE_SENTINEL" = "1" ] && [ ! -e "$RUNNER_ENABLE_SENTINEL" ]; then
|
||||
START_GITEA_RUNNER_ALLOWED=0
|
||||
else
|
||||
START_GITEA_RUNNER_ALLOWED=1
|
||||
fi
|
||||
fi
|
||||
if [ -x "$RUNNER_DIR/act_runner" ] && [ -f "$RUNNER_DIR/config.yaml" ]; then
|
||||
# 若舊的 .runner 配置指向過期 hostname，只有在明確允許啟動 runner
|
||||
@@ -263,14 +271,19 @@ PY
|
||||
nohup "$RUNNER_DIR/run-host-runner.sh" >> "$RUNNER_DIR/host-runner.log" 2>&1 &
|
||||
fi
|
||||
else
|
||||
if [ "$START_GITEA_RUNNER_ON_BOOT" = "1" ]; then
|
||||
log "⛔ AWOOOI_START_GITEA_RUNNER_ON_BOOT=1 但缺少 $RUNNER_ENABLE_SENTINEL；runner fail-closed"
|
||||
if [ "$START_GITEA_RUNNER_ON_BOOT" = "1" ] && [ "$REQUIRE_RUNNER_ENABLE_SENTINEL" = "1" ]; then
|
||||
log "⛔ AWOOOI_START_GITEA_RUNNER_ON_BOOT=1 但缺少 $RUNNER_ENABLE_SENTINEL；runner startup 暫停"
|
||||
else
|
||||
log "⏸️ Gitea host runner 維持停用；需同時設定 AWOOOI_START_GITEA_RUNNER_ON_BOOT=1 與建立 $RUNNER_ENABLE_SENTINEL 才允許 startup 啟動"
|
||||
log "⏸️ Gitea host runner 本次不啟動；AWOOOI_START_GITEA_RUNNER_ON_BOOT=1 可重新打開"
|
||||
fi
|
||||
if [ "$STOP_GITEA_RUNNER_WHEN_DISABLED" = "1" ]; then
|
||||
log "⚠️ AWOOI_STOP_GITEA_RUNNER_WHEN_DISABLED=1，停止 runner"
|
||||
systemctl disable --now "$RUNNER_SERVICE" >/dev/null 2>&1 || true
|
||||
systemctl kill -s SIGKILL "$RUNNER_SERVICE" >/dev/null 2>&1 || true
|
||||
pkill -KILL -f "$RUNNER_DIR/act_runner daemon" >/dev/null 2>&1 || true
|
||||
else
|
||||
log "✅ 不停止既有 runner；避免中斷正在執行的 CD / post-deploy job"
|
||||
fi
|
||||
systemctl disable --now "$RUNNER_SERVICE" >/dev/null 2>&1 || true
|
||||
systemctl kill -s SIGKILL "$RUNNER_SERVICE" >/dev/null 2>&1 || true
|
||||
pkill -KILL -f "$RUNNER_DIR/act_runner daemon" >/dev/null 2>&1 || true
|
||||
fi
|
||||
|
||||
# 已停用 Docker-wrapped runner；避免它搶走 host label job。
|
||||
|
||||
Reference in New Issue
Block a user