[danielvm-git]

User description

Summary

Fixes BUG-2026-11-06-dashboard-metrics-hang — the v2.93.0 regression that left the admin dashboard stuck on "Loading dashboard..." ("the page wasn't loading") and forced the earlier rollback.

Root cause

06f86d39d made /api/monitoring/metrics run the new state-age report against deployments + cici_runs using the unbounded request context. The DashboardPage awaits that endpoint before setLoading(false), so a slow or write-locked DB left the page loading forever.

Fix

Apply the existing stateAgeQueryTimeout (2s) to both metrics endpoints — rename stateAgeReportForAlerts → stateAgeReportTimed and use it in handleMetrics and handlePrometheusMetrics.

Test plan

• go test -run TestStateAgeReportTimedBoundsBlockedQuery ./components/monitoring — PASS (blocks QueryContext, asserts bounded return)
• go test ./components/monitoring — PASS
• go test -race ./components/monitoring — PASS
• go test ./... — PASS
• Preflight (go vet + go test + gosec + build + codeql + gitleaks) — green
• trace-stories --strict — exit 0; 24 gate-script tests pass

Note

Merging is a fix: commit, so big-release will publish a patch (v2.94.1) and the Deploy workflow will fire. That deploy is still blocked by the separate VPS-side SCP issue (see #312), tracked independently of this fix.

---

PR Type

Bug fix, Tests

---

Description

• Bound state-age DB queries in metrics endpoints with 2s timeout

• Replace unbounded request-context call with stateAgeReportTimed

• Add regression test blocking QueryContext to assert bounded return

• Prevent admin dashboard hang caused by v2.93.0 regression

---

Diagram Walkthrough

flowchart LR
  A["handleMetrics"] -- "uses" --> B["stateAgeReportTimed"]
  C["handlePrometheusMetrics"] -- "uses" --> B
  D["stateAgeMetricValue"] -- "uses" --> B
  B -- "context.WithTimeout(stateAgeQueryTimeout)" --> E["bounded DB query"]
  E -- "returns unavailable report promptly" --> F["no dashboard hang"]

Loading

File Walkthrough

	Relevant files
Bug fix	monitoring.go Switch metrics endpoints to timed state-age report              components/monitoring/monitoring.go Update handleMetrics to call m.stateAgeReportTimed() instead of m.stateAgeReport(r.Context()) Update handlePrometheusMetrics to call m.stateAgeReportTimed() in writeStateAgeGauges +2/-2      state_age.go Rename and document bounded state-age report                          components/monitoring/state_age.go Rename stateAgeReportForAlerts to stateAgeReportTimed Add comment explaining bounded query prevents dashboard hang Update stateAgeMetricValue to call renamed function +7/-3
Tests	state_age_test.go Add regression test for blocked state-age query                    components/monitoring/state_age_test.go Add blockingDBer test double that blocks QueryContext until context cancellation Add TestStateAgeReportTimedBoundsBlockedQuery to verify bounded query in blocked scenario Assert report returns within timeout and marks all state machines unavailable +52/-0

---