docs: clarify health dashboard remediation (#35)

This commit is contained in:
2026-05-10 08:57:31 +01:00
committed by GitHub
parent 5d3ec80907
commit 796d4ae381
+34 -8
View File
@@ -5,18 +5,44 @@ Use this runbook when the repository supervisor health dashboard becomes stale.
## Signals ## Signals
- The pinned supervisor issue has an old `last_refresh:` marker. - The pinned supervisor issue has an old `last_refresh:` marker.
- The stats scheduler exists and is loaded, but wrapper runs overlap or remain active beyond the expected timeout. - The stats scheduler exists and is loaded, but wrapper runs overlap or remain
active beyond the expected timeout.
- Recent stats logs stop before the affected repository is updated. - Recent stats logs stop before the affected repository is updated.
## Remediation ## Remediation
1. Confirm the launchd job and stats log are present. 1. Confirm the scheduler and stats log are present:
```bash
launchctl list | grep -i aidevops-stats-wrapper
ls -la ~/Library/LaunchAgents/com.aidevops.aidevops-stats-wrapper.plist
tail -40 ~/.aidevops/logs/stats.log
```
2. Check for an active `stats-wrapper.sh` process that has exceeded `STATS_TIMEOUT`. 2. Check for an active `stats-wrapper.sh` process that has exceeded `STATS_TIMEOUT`.
3. Terminate the stale wrapper process, remove the stale stats pidfile, and run one targeted health issue refresh for this repository. The default timeout is `600` seconds and is defined near the top of
4. Verify the pinned dashboard issue now has a fresh `last_refresh:` marker and recent `updated_at` timestamp. `~/.aidevops/agents/scripts/stats-wrapper.sh`; an operator can confirm the
runtime value with a dry run:
## Verification evidence for issue #32 ```bash
STATS_DRY_RUN=1 bash ~/.aidevops/agents/scripts/stats-wrapper.sh --dry-run
```
- Scheduler plist existed and launchd reported `com.aidevops.aidevops-stats-wrapper` loaded. 3. Terminate the stale wrapper process and remove the stale stats pidfile at
- The stats log showed repeated overlapping runs and a stale wrapper process, with the repository dashboard skipped before the targeted refresh. `~/.aidevops/logs/stats.pid`.
- A targeted refresh updated dashboard issue #10 with `last_refresh: 2026-05-10T00:38:30Z`.
4. Run one targeted health issue refresh for this repository:
```bash
REPO_SLUG="wpallstars/wp-fix-plugin-does-not-exist-notices" \
REPO_PATH="$HOME/Git/wordpress/wp-fix-plugin-does-not-exist-notices" \
bash -lc '
source "$HOME/.aidevops/agents/scripts/shared-constants.sh"
source "$HOME/.aidevops/agents/scripts/worker-lifecycle-common.sh"
source "$HOME/.aidevops/agents/scripts/stats-functions.sh"
_update_health_issue_for_repo "$REPO_SLUG" "$REPO_PATH" "" "" ""
'
```
5. Verify the pinned dashboard issue now has a fresh `last_refresh:` marker and
recent `updated_at` timestamp.