grafana/k8s-monitoring-helm k8s-monitoring-4.2.2

on GitHub

What's changed

• Add attachNamespaceMetadata option to the Pod Logs (via Loki and via Kubernetes API) features. When enabled, namespace labels are attached to pod discovery targets and made available as __meta_kubernetes_namespace_label_*. (@petewall, @tdudas)
• Fix Beyla auto-instrumentation discarding the default discovery targets (instrument/exclude_instrument) when a user set any other field under discovery, such as exclude_otel_instrumented_services. The defaults are now applied per-field, so a partial discovery config no longer stops Beyla from discovering services. (#2332) (@petewall)
• Add host-tracefs collector preset.
• Remove empty image and container labels from cAdvisor's container_pressure_.* metrics (#2561) (@petewall)
• Infer auth.type: basic for remote configuration and destinations when a username and password are set but auth.type is not. (#2809) (@TylerHelmuth)
• Fix unrecognized block name "rule_namespace" error when using the rules.namespaces field of the prometheus destination to limit the PrometheusRules object discovery to specific namespaces (#2802) (@sebasnabas)
• Fix autoInstrumentation.beyla.service.targetPort not propagating to the generated Beyla config ports, which previously required setting the port in three places. (#2811) (@TylerHelmuth)
• Update Beyla to 1.16.10 (@TylerHelmuth)
• Update Alloy Operator to 0.6.1, kube-state-metrics to 7.8.1, Node Exporter to 4.56.0, and OpenCost to 2.5.26 (@petewall)* Use journald priority as a fallback log level when the log message does not contain a log level. (#2827) (@saruprim)

Chart released at https://github.com/grafana/helm-charts/releases/tag/k8s-monitoring-4.2.2

Full changelog: CHANGELOG.md