Skip to main content

Monitoring & Alerts (Experience Plugins)

Metrics (suggested)​

  • plugin_bundle_serve_ms{tenant,plugin_id,version} (histogram)
  • plugin_allowlist_violations_total{tenant,plugin_id} (counter)
  • plugin_quarantine_status{tenant,plugin_id} (gauge 0/1)

PromQL examples​

  • p95 bundle serve latency per plugin:
histogram_quantile(0.95, sum(rate(plugin_bundle_serve_ms_bucket[5m])) by (le, plugin_id))
  • Violation rate (> 5/min):
sum(rate(plugin_allowlist_violations_total[1m])) by (plugin_id) > 5
  • Any quarantine active:
max_over_time(plugin_quarantine_status[5m]) == 1

LogQL (Loki)​

{app="bff", level="warn"} |= "X-Allowlist-Violation" | unwrap plugin_id | count_over_time([5m])

Alerts (suggested)​

  • Violation spike: warn at >5/min, crit at >20/min (per plugin)
  • Latency: p95 bundle serve > 500ms sustained 10m
  • Quarantine: any plugin in quarantine (notify ops immediately)

See also: Ops Runbook ./ops-runbook, CI/CD Cookbook ./cicd-cookbook.