Repository navigation
Commit 5fc5b67
feat(alerting): convert 3 hand-authored VMRules to Grafana alert rules (#559) (#24)
Adds terraform/alerting-converted-vmrules.tf with grafana_rule_group
resources reproducing the three VMRule CRDs that today are vmalert-evaluated
with no non-paging delivery path:
- keycloak-dr-backup -> KeycloakRealmBackupStale (severity=ticket)
- tailscale-operator -> TailscaleOperatorDown / TailscaleProxyCrashLoop
/ TailscaleOAuthSecretSyncErrors (severity=warning)
- cert-expiry-euc1 -> EUC1CertExpiringIn14Days / EUC1CertNotReady
(severity=ticket)
Each rule runs the original PromQL expr against the VictoriaMetrics
datasource (P4169E866C3094E38) in the A step, reduces to last in B, and
fires on C threshold > 0 (the source exprs already bake the comparison into
a boolean-ish series). Pattern matches the applied alerting-euc1-dr.tf.
severity=warning/ticket labels route via the EXISTING non-paging policy
(pagerduty-platform-infra-low -> low-urgency PD -> Slack #mzla-pages). The
notification policy is not modified.
The paired platform-infrastructure PR removing the VMRule YAMLs must merge
AFTER these rules are applied, to avoid an alerting gap.
Relates: thunderbird/platform-infrastructure#559
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>1 parent 3c08fae commit 5fc5b67
1 file changed
Lines changed: 491 additions & 0 deletions
0 commit comments