fix(reboot): enforce direct runner fail-closed guard [skip ci]
This commit is contained in:
@@ -20,6 +20,25 @@
|
||||
- OpenClaw 仍維持 production decision core;替換前必須 replay / shadow / canary / ADR。
|
||||
- SDK install、API shadow / canary、production route、paid provider / cost route、external active security scan、secret value / credential URL / raw env、DB destructive / backup restore、force push / repo refs deletion 仍不得被本段 controlled queue 直接打開。
|
||||
|
||||
## 2026-06-28 — 09:16 direct runner source guard 實作收斂
|
||||
|
||||
**背景**:09:00 前的 live hotfix 已把 110 上 direct / Gitea runner 全部 mask,但 `awoooi-startup-110.sh`、cold-start 與 P3 release gate 還沒有把 `awoooi-direct-runner-open.service` 這條 transient direct runner 路徑納入 source-level guard。
|
||||
|
||||
**完成內容**:
|
||||
- `scripts/reboot-recovery/awoooi-startup-110.sh` 新增 `RUNNER_FAIL_CLOSED_SERVICES` 與 `RUNNER_FAIL_CLOSED_BINARY_PATHS`,預設未同時具備 `AWOOOI_START_GITEA_RUNNER_ON_BOOT=1` 與 `/run/awoooi-runner-host-enabled` 時,會強制 kill / disable / mask direct runner 與 Gitea runner units,並把 live runner ELF quarantine 成 163-byte fail-closed stub。
|
||||
- `scripts/reboot-recovery/full-stack-cold-start-check.sh` 新增 110 runner fail-closed readback:direct / Gitea units 必須 `load=masked unitfile=masked active=inactive`,direct runner process count 必須 `0`,runner binary 不得是 ELF。
|
||||
- `scripts/reboot-recovery/post-start-quick-check.sh` 新增 `110 runner fail-closed guard` section,並以 `HOST_WEB_BUILD_PRESSURE_ATTEMPTS=1` 讀回 pressure gate。
|
||||
- `scripts/reboot-recovery/p3-controlled-release-gate.sh` 將 direct runner fail-closed 狀態納入 `BAD_RUNNER_GUARDRAILS`,避免 P3 release gate 只看 `actions.runner.*` 而漏掉 transient direct runner。
|
||||
- Live `/usr/local/bin/awoooi-startup-110.sh` 已更新並加 immutable;讀回 `LIVE_STARTUP_DIRECT_UNIT=1`、`LIVE_STARTUP_GUARD_FUNC=2`、`LIVE_STARTUP_DEFAULT=failclosed`。
|
||||
|
||||
**驗證結果**:
|
||||
- 本地:`bash -n` 通過 `awoooi-startup-110.sh`、`full-stack-cold-start-check.sh`、`post-start-quick-check.sh`、`p3-controlled-release-gate.sh`;`git diff --check` 通過;direct runner source invariant 通過。
|
||||
- quick-check runner-only:`POST_START_QUICK_CHECK PASS=13 WARN=0 BLOCKED=0`、`RESULT=GREEN`;六個 runner/direct units 全部 masked / inactive、runner process `0`、四條 binary path 皆為 shell stub、pressure gate `RUNNER_PRESSURE_GATE_RC 0`。
|
||||
- cold-start 單次讀回:runner guard OK;整體仍 `PASS=90 WARN=1 BLOCKED=1`、`Result: BLOCKED`,blocker 是 `188 momo daily sales data stale beyond 3 days`,不是 runner。
|
||||
- P3 release gate:runner/CD guardrails 顯示 `BAD_RUNNER_GUARDRAILS 0`;整體仍 `HOLD_P3_RELEASE`,blockers 包含 cold-start、188 backup stale、188 litellm not running。
|
||||
|
||||
**邊界**:本段沒有重啟 Docker / Nginx / firewall / K3s / DB,沒有讀 raw sessions / SQLite / auth / `.env` / runner token,也沒有恢復 110 runner。
|
||||
|
||||
## 2026-06-28 — 08:45 110 runner 壓力事故 source / live fail-closed 收斂
|
||||
|
||||
**背景**:統帥全面授權打開非事故級 gate,但 110 Gitea runner 反覆拉起 StockPlatform headless Chrome smoke,已造成 production host CPU / CI 壓力事故;runner 未搬遷 / 限流前不得直接重開。
|
||||
|
||||
@@ -194,11 +194,105 @@ RUNNER_SERVICE="gitea-act-runner-host.service"
|
||||
RUNNER_ENABLE_SENTINEL="/run/awoooi-runner-host-enabled"
|
||||
START_GITEA_RUNNER_ON_BOOT="${AWOOOI_START_GITEA_RUNNER_ON_BOOT:-0}"
|
||||
START_GITEA_RUNNER_ALLOWED=0
|
||||
RUNNER_FAIL_CLOSED_SERVICES=(
|
||||
"awoooi-direct-runner-open.service"
|
||||
"awoooi-direct-runner.service"
|
||||
"gitea-act-runner-host.service"
|
||||
"gitea-act-runner-awoooi-controlled.service"
|
||||
"gitea-awoooi-controlled-runner.service"
|
||||
"gitea-act-runner-awoooi-open.service"
|
||||
)
|
||||
RUNNER_FAIL_CLOSED_BINARY_PATHS=(
|
||||
"/home/wooo/act-runner/act_runner"
|
||||
"/home/wooo/act-runner/act_runner.real-20260628-runner-pressure-guard"
|
||||
"/home/wooo/act-runner-controlled/act_runner"
|
||||
"/home/wooo/awoooi-controlled-runner/awoooi_controlled_runner"
|
||||
)
|
||||
# The runtime operator sentinel is the second key for an authorized deployment
|
||||
# window. A single env var or a stale sentinel alone must not reopen host CI.
|
||||
if [ "$START_GITEA_RUNNER_ON_BOOT" = "1" ] && [ -e "$RUNNER_ENABLE_SENTINEL" ]; then
|
||||
START_GITEA_RUNNER_ALLOWED=1
|
||||
fi
|
||||
|
||||
mask_runner_unit_file() {
|
||||
local unit="$1"
|
||||
local unit_dir="$2"
|
||||
local owner_user="${3:-}"
|
||||
local unit_file="$unit_dir/$unit"
|
||||
local quarantine_stamp
|
||||
quarantine_stamp="$(date +%Y%m%d%H%M%S)"
|
||||
|
||||
mkdir -p "$unit_dir" >/dev/null 2>&1 || true
|
||||
if [ -L "$unit_file" ] && [ "$(readlink "$unit_file" 2>/dev/null || true)" = "/dev/null" ]; then
|
||||
return 0
|
||||
fi
|
||||
if [ -e "$unit_file" ] || [ -L "$unit_file" ]; then
|
||||
chattr -i "$unit_file" >/dev/null 2>&1 || true
|
||||
mv "$unit_file" "${unit_file}.quarantined-runner-incident-${quarantine_stamp}" >/dev/null 2>&1 || true
|
||||
fi
|
||||
ln -s /dev/null "$unit_file" >/dev/null 2>&1 || true
|
||||
if [ -n "$owner_user" ]; then
|
||||
chown -h "$owner_user:$owner_user" "$unit_file" >/dev/null 2>&1 || true
|
||||
fi
|
||||
}
|
||||
|
||||
guard_runner_binary_fail_closed() {
|
||||
local path="$1"
|
||||
local tmp
|
||||
local quarantine_stamp
|
||||
quarantine_stamp="$(date +%Y%m%d%H%M%S)"
|
||||
|
||||
if [ -e "$path" ]; then
|
||||
chattr -i "$path" >/dev/null 2>&1 || true
|
||||
if file "$path" 2>/dev/null | grep -qi "ELF"; then
|
||||
mv "$path" "${path}.quarantined-runner-incident-${quarantine_stamp}" >/dev/null 2>&1 || true
|
||||
chmod 0400 "${path}.quarantined-runner-incident-${quarantine_stamp}" >/dev/null 2>&1 || true
|
||||
chattr +i "${path}.quarantined-runner-incident-${quarantine_stamp}" >/dev/null 2>&1 || true
|
||||
fi
|
||||
fi
|
||||
|
||||
tmp="$(mktemp)"
|
||||
cat >"$tmp" <<'EOF'
|
||||
#!/usr/bin/env bash
|
||||
set -eu
|
||||
echo "AWOOOI host runner is fail-closed on 110 after 2026-06-28 pressure incident; migrate or rate-limit before enabling." >&2
|
||||
exit 75
|
||||
EOF
|
||||
install -o root -g root -m 0755 "$tmp" "$path" >/dev/null 2>&1 || true
|
||||
rm -f "$tmp"
|
||||
chattr +i "$path" >/dev/null 2>&1 || true
|
||||
}
|
||||
|
||||
ensure_host_runner_fail_closed() {
|
||||
local unit
|
||||
local binary
|
||||
local wooo_uid
|
||||
|
||||
for unit in "${RUNNER_FAIL_CLOSED_SERVICES[@]}"; do
|
||||
systemctl kill --signal=SIGKILL "$unit" >/dev/null 2>&1 || true
|
||||
systemctl reset-failed "$unit" >/dev/null 2>&1 || true
|
||||
systemctl disable "$unit" >/dev/null 2>&1 || true
|
||||
systemctl mask "$unit" >/dev/null 2>&1 || mask_runner_unit_file "$unit" "/etc/systemd/system"
|
||||
mask_runner_unit_file "$unit" "/etc/systemd/system"
|
||||
done
|
||||
systemctl daemon-reload >/dev/null 2>&1 || true
|
||||
|
||||
if wooo_uid="$(id -u wooo 2>/dev/null)"; then
|
||||
mkdir -p /home/wooo/.config/systemd/user >/dev/null 2>&1 || true
|
||||
for unit in "${RUNNER_FAIL_CLOSED_SERVICES[@]}"; do
|
||||
if [ -d "/run/user/$wooo_uid" ] && command -v runuser >/dev/null 2>&1; then
|
||||
runuser -u wooo -- env XDG_RUNTIME_DIR="/run/user/$wooo_uid" systemctl --user kill --signal=SIGKILL "$unit" >/dev/null 2>&1 || true
|
||||
fi
|
||||
mask_runner_unit_file "$unit" "/home/wooo/.config/systemd/user" "wooo"
|
||||
done
|
||||
fi
|
||||
|
||||
pkill -KILL -f "^${RUNNER_DIR}/act_runner(\\.real-[^ ]*)? daemon" >/dev/null 2>&1 || true
|
||||
for binary in "${RUNNER_FAIL_CLOSED_BINARY_PATHS[@]}"; do
|
||||
guard_runner_binary_fail_closed "$binary"
|
||||
done
|
||||
}
|
||||
|
||||
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
|
||||
# 時才清除重新註冊;預設降壓模式不得碰 registration 狀態。
|
||||
@@ -271,9 +365,7 @@ PY
|
||||
else
|
||||
log "⏸️ Gitea host runner 維持停用;需同時設定 AWOOOI_START_GITEA_RUNNER_ON_BOOT=1 與建立 $RUNNER_ENABLE_SENTINEL 才允許 startup 啟動"
|
||||
fi
|
||||
systemctl disable --now "$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
|
||||
ensure_host_runner_fail_closed
|
||||
fi
|
||||
|
||||
# 已停用 Docker-wrapped runner;避免它搶走 host label job。
|
||||
|
||||
@@ -286,6 +286,20 @@ echo "ACTION_RUNNER_ENABLED_COUNT $(systemctl list-unit-files "actions.runner.*"
|
||||
for u in $(systemctl list-units "actions.runner.*" --all --no-legend --plain 2>/dev/null | awk "{print \$1}"); do
|
||||
systemctl show "$u" -p ActiveState -p SubState -p CPUQuotaPerSecUSec -p MemoryMax -p WatchdogUSec -p NRestarts | sed "s/^/RUNNER $u /"
|
||||
done
|
||||
for u in awoooi-direct-runner-open.service awoooi-direct-runner.service gitea-act-runner-host.service gitea-act-runner-awoooi-controlled.service gitea-awoooi-controlled-runner.service gitea-act-runner-awoooi-open.service; do
|
||||
load=$(systemctl show "$u" -p LoadState --value 2>/dev/null || true)
|
||||
unitfile=$(systemctl show "$u" -p UnitFileState --value 2>/dev/null || true)
|
||||
active=$(systemctl show "$u" -p ActiveState --value 2>/dev/null || true)
|
||||
mainpid=$(systemctl show "$u" -p MainPID --value 2>/dev/null || true)
|
||||
echo "RUNNER_FAILCLOSED_UNIT $u load=$load unitfile=$unitfile active=$active mainpid=$mainpid"
|
||||
done
|
||||
direct_runner_count=$(pgrep -f "^/home/wooo/act-runner/act_runner|^/home/wooo/act-runner-controlled/act_runner|^/home/wooo/awoooi-controlled-runner/awoooi_controlled_runner" 2>/dev/null | wc -l | tr -d " ")
|
||||
echo "RUNNER_DIRECT_PROCESS_COUNT $direct_runner_count"
|
||||
for p in /home/wooo/act-runner/act_runner /home/wooo/act-runner/act_runner.real-20260628-runner-pressure-guard /home/wooo/act-runner-controlled/act_runner /home/wooo/awoooi-controlled-runner/awoooi_controlled_runner; do
|
||||
kind=$(file -b "$p" 2>/dev/null || echo missing)
|
||||
echo "RUNNER_FAILCLOSED_BINARY $p kind=$kind"
|
||||
echo "$kind" | grep -qi "ELF" && echo "RUNNER_FAILCLOSED_BINARY_ELF $p"
|
||||
done
|
||||
docker ps --format "DOCKER {{.Names}}\t{{.Status}}" | head -120
|
||||
' 2>&1); then
|
||||
fail "ssh 110 read-only check"
|
||||
@@ -309,6 +323,13 @@ docker ps --format "DOCKER {{.Names}}\t{{.Status}}" | head -120
|
||||
else
|
||||
warn "runner watchdog state not confirmed"
|
||||
fi
|
||||
if awk '$1 == "RUNNER_FAILCLOSED_UNIT" && ($3 != "load=masked" || $4 != "unitfile=masked") {bad=1} END {exit bad}' <<<"$out"; then
|
||||
ok "110 direct/Gitea runner fail-closed units are masked"
|
||||
else
|
||||
fail "110 direct/Gitea runner fail-closed units are not all masked"
|
||||
fi
|
||||
grep -q "RUNNER_DIRECT_PROCESS_COUNT 0" <<<"$out" && ok "110 direct runner process count is zero" || fail "110 direct runner process detected"
|
||||
grep -q "RUNNER_FAILCLOSED_BINARY_ELF" <<<"$out" && fail "110 runner fail-closed binary path restored to ELF" || ok "110 runner binary paths are fail-closed stubs or missing"
|
||||
grep -q "sentry-self-hosted-clickhouse-1.*Restarting" <<<"$out" && warn "Sentry ClickHouse restarting" || ok "Sentry ClickHouse not visibly restarting"
|
||||
}
|
||||
|
||||
|
||||
@@ -304,8 +304,23 @@ awk '
|
||||
check_runner_guardrails() {
|
||||
section "runner/CD guardrails"
|
||||
local out bad
|
||||
if ! out=$(ssh_cmd "wooo@192.168.0.110" '
|
||||
if ! out=$(ssh_cmd "wooo@192.168.0.110" '
|
||||
bad=0
|
||||
for u in awoooi-direct-runner-open.service awoooi-direct-runner.service gitea-act-runner-host.service gitea-act-runner-awoooi-controlled.service gitea-awoooi-controlled-runner.service gitea-act-runner-awoooi-open.service; do
|
||||
load=$(systemctl show "$u" -p LoadState --value 2>/dev/null || true)
|
||||
unitfile=$(systemctl show "$u" -p UnitFileState --value 2>/dev/null || true)
|
||||
active=$(systemctl show "$u" -p ActiveState --value 2>/dev/null || true)
|
||||
echo "RUNNER_FAILCLOSED_UNIT $u load=$load unitfile=$unitfile active=$active"
|
||||
[ "$load" = "masked" ] && [ "$unitfile" = "masked" ] && [ "$active" = "inactive" ] || bad=1
|
||||
done
|
||||
direct_runner_count=$(pgrep -f "^/home/wooo/act-runner/act_runner|^/home/wooo/act-runner-controlled/act_runner|^/home/wooo/awoooi-controlled-runner/awoooi_controlled_runner" 2>/dev/null | wc -l | tr -d " ")
|
||||
echo "RUNNER_DIRECT_PROCESS_COUNT $direct_runner_count"
|
||||
[ "$direct_runner_count" = "0" ] || bad=1
|
||||
for p in /home/wooo/act-runner/act_runner /home/wooo/act-runner/act_runner.real-20260628-runner-pressure-guard /home/wooo/act-runner-controlled/act_runner /home/wooo/awoooi-controlled-runner/awoooi_controlled_runner; do
|
||||
kind=$(file -b "$p" 2>/dev/null || echo missing)
|
||||
echo "RUNNER_FAILCLOSED_BINARY $p kind=$kind"
|
||||
echo "$kind" | grep -qi "ELF" && bad=1
|
||||
done
|
||||
for u in $(systemctl list-units "actions.runner.*" --all --no-legend --plain 2>/dev/null | awk "{print \$1}"); do
|
||||
watchdog=$(systemctl show "$u" -p WatchdogUSec --value)
|
||||
quota=$(systemctl show "$u" -p CPUQuotaPerSecUSec --value)
|
||||
|
||||
@@ -535,6 +535,41 @@ if [[ "$RUN_CPU" -eq 1 ]]; then
|
||||
rm -f "$cpu_tmp"
|
||||
fi
|
||||
|
||||
section "110 runner fail-closed guard"
|
||||
runner_tmp="$(mktemp -t post-start-runner.XXXXXX)"
|
||||
if ssh_read "wooo@192.168.0.110" '
|
||||
for u in awoooi-direct-runner-open.service awoooi-direct-runner.service gitea-act-runner-host.service gitea-act-runner-awoooi-controlled.service gitea-awoooi-controlled-runner.service gitea-act-runner-awoooi-open.service; do
|
||||
load=$(systemctl show "$u" -p LoadState --value 2>/dev/null || true)
|
||||
unitfile=$(systemctl show "$u" -p UnitFileState --value 2>/dev/null || true)
|
||||
active=$(systemctl show "$u" -p ActiveState --value 2>/dev/null || true)
|
||||
mainpid=$(systemctl show "$u" -p MainPID --value 2>/dev/null || true)
|
||||
echo "RUNNER_FAILCLOSED_UNIT $u load=$load unitfile=$unitfile active=$active mainpid=$mainpid"
|
||||
done
|
||||
direct_runner_count=$(pgrep -f "^/home/wooo/act-runner/act_runner|^/home/wooo/act-runner-controlled/act_runner|^/home/wooo/awoooi-controlled-runner/awoooi_controlled_runner" 2>/dev/null | wc -l | tr -d " ")
|
||||
echo "RUNNER_DIRECT_PROCESS_COUNT $direct_runner_count"
|
||||
for p in /home/wooo/act-runner/act_runner /home/wooo/act-runner/act_runner.real-20260628-runner-pressure-guard /home/wooo/act-runner-controlled/act_runner /home/wooo/awoooi-controlled-runner/awoooi_controlled_runner; do
|
||||
kind=$(file -b "$p" 2>/dev/null || echo missing)
|
||||
echo "RUNNER_FAILCLOSED_BINARY $p kind=$kind"
|
||||
echo "$kind" | grep -qi "ELF" && echo "RUNNER_FAILCLOSED_BINARY_ELF $p"
|
||||
done
|
||||
HOST_WEB_BUILD_PRESSURE_ATTEMPTS=1 HOST_WEB_BUILD_PRESSURE_SLEEP_SECONDS=0 /usr/local/bin/awoooi-wait-host-web-build-pressure.sh
|
||||
echo "RUNNER_PRESSURE_GATE_RC $?"
|
||||
' >"$runner_tmp" 2>&1; then
|
||||
ok "110 runner fail-closed readback succeeded"
|
||||
else
|
||||
blocked "110 runner fail-closed readback failed"
|
||||
fi
|
||||
cat "$runner_tmp"
|
||||
if awk '$1 == "RUNNER_FAILCLOSED_UNIT" && ($3 != "load=masked" || $4 != "unitfile=masked") {bad=1} END {exit bad}' "$runner_tmp"; then
|
||||
ok "110 direct/Gitea runner fail-closed units are masked"
|
||||
else
|
||||
blocked "110 direct/Gitea runner fail-closed units are not all masked"
|
||||
fi
|
||||
grep -q "RUNNER_DIRECT_PROCESS_COUNT 0" "$runner_tmp" && ok "110 direct runner process count is zero" || blocked "110 direct runner process detected"
|
||||
grep -q "RUNNER_FAILCLOSED_BINARY_ELF" "$runner_tmp" && blocked "110 runner fail-closed binary path restored to ELF" || ok "110 runner binary paths are fail-closed stubs or missing"
|
||||
grep -q "RUNNER_PRESSURE_GATE_RC 0" "$runner_tmp" && ok "110 host pressure gate returned 0" || blocked "110 host pressure gate is blocking"
|
||||
rm -f "$runner_tmp"
|
||||
|
||||
section "總結"
|
||||
printf 'POST_START_QUICK_CHECK PASS=%s WARN=%s BLOCKED=%s\n' "$PASS_COUNT" "$WARN_COUNT" "$BLOCKED_COUNT"
|
||||
printf 'POST_START_QUICK_CHECK_WARNINGS SERVICE=%s BOUNDARY=%s EVIDENCE=%s\n' "$SERVICE_WARN_COUNT" "$BOUNDARY_WARN_COUNT" "$EVIDENCE_WARN_COUNT"
|
||||
|
||||
Reference in New Issue
Block a user