[WHM] Metrics Collector - 3 Consecutive Failures, Metrics Infrastructure Degraded #23446
Closed as not planned
Description
Workflow: Metrics Collector - Infrastructure Agent
Priority: P1 — Metrics infrastructure is degraded, affecting all meta-orchestrators.
Status
| Run | Date | Conclusion |
|---|---|---|
| #108 (23692639793) | 2026-03-28T19:29Z | ❌ failure |
| #107 (23663993174) | 2026-03-27T19:32Z | ❌ failure |
| #106 (23614384478) | 2026-03-26T19:39Z | ❌ failure |
| #105 (23560285208) | 2026-03-25T19:35Z | ✅ success ← last success |
| #104 (23508768948) | 2026-03-24T19:38Z | ✅ success |
Metrics are stale since 2026-03-25 (last successful run was #105).
Root Cause (from WHM analysis)
The agent is writing agent-performance-latest.md to the artifact root instead of metrics/latest.json. The push_repo_memory job uses a glob filter metrics/** which rejects root-level files → 0 metrics are persisted.
Secondary issue: The push_repo_memory → Post Setup Scripts systemic bug (same as #23151) causes the workflow to fail in cleanup even when the agent itself may have succeeded.
Impact
/tmp/gh-aw/repo-memory/default/metrics/latest.jsonstale since 2026-03-25- Workflow Health Manager cannot query execution metrics
- Agent Performance Analyzer metrics unavailable
- Health scoring is less accurate (relying on GitHub API calls instead)
Suggested Fix
- Check
metrics-collector.mdworkflow prompt — ensure the agent is instructed to write tometrics/latest.json(not root-level files) - Verify the
push_repo_memoryglob pattern matches the output file paths - Fix the systemic
push_repo_memory → Post Setup Scriptsbug (see [aw] PR Triage Agent failed #23151)
References
- Detected by WHM run §23708489882
- Related systemic issue: #23151
Generated by Workflow Health Manager - Meta-Orchestrator
Generated by Workflow Health Manager - Meta-Orchestrator · ◷
- expires on Mar 30, 2026, 12:04 PM UTC