Release 0.5.35 — Prometheus ServiceMonitor Fix and OpenFGA Metrics
Released: 2026-07-02 Chart:
oci://ghcr.io/cnoe-io/charts/ai-platform-engineering:0.5.35Previous release: 0.5.34
Highlights
0.5.35 resolves persistent TargetDown Prometheus alerts for agentgateway, keycloak, openfga, and audit-service by fixing label-selector gaps in the umbrella ServiceMonitor and adding a dedicated OpenFGA metrics ServiceMonitor.
Bug Fixes
Prometheus umbrella ServiceMonitor no longer scrapes label-less services
The umbrella servicemonitor.yaml used a NotIn label selector on app.kubernetes.io/component to exclude per-component services. However, agentgateway, keycloak, openfga, and audit-service lack that label entirely — Kubernetes evaluates a NotIn selector on a missing key as true, so these services were included and produced permanently-failing TargetDown targets (their /metrics endpoints are not on the umbrella SM's http port).
Fixed by adding a matching NotIn exclusion on app.kubernetes.io/name for these four services. Each service that has a metrics endpoint now has its own dedicated ServiceMonitor:
| Service | ServiceMonitor | Port |
|---|---|---|
agentgateway | servicemonitor-agentgateway.yaml | 15000 (admin) |
keycloak | servicemonitor-keycloak.yaml | 9000 (management) |
openfga | servicemonitor-openfga.yaml | 2112 (metrics) |
audit-service | — (no metrics endpoint) | — |
(#2106)
OpenFGA Prometheus metrics port exposed
The OpenFGA deployment and service now expose the built-in Prometheus metrics port (2112), which OpenFGA serves by default. The new dedicated servicemonitor-openfga.yaml targets this port instead of the API port (8080). (#2106)
Breaking Changes
No breaking changes. Drop-in upgrade from 0.5.34.
Upgrade Guide: 0.5.34 → 0.5.35
No values changes required. After upgrading, verify Prometheus targets:
helm upgrade ai-platform-engineering \
oci://ghcr.io/cnoe-io/charts/ai-platform-engineering \
--version 0.5.35 \
-f your-values.yaml
Verify the four dedicated ServiceMonitors are created:
kubectl get servicemonitor -n <namespace>
# expect: caipe-metrics, caipe-agentgateway-metrics, caipe-keycloak-metrics, caipe-openfga-metrics
