grafana/k8s-monitoring-helm k8s-monitoring-4.1.4

on GitHub

What's changed

• Fix the Service Integrations feature silently rendering no integration modules when integrations.collector was left at its default empty string. The feature now relies on the same collector resolution as every other feature, so simply enabling integrations (e.g. integrations.alloy, integrations.cert-manager, integrations.istio) generates the expected Alloy modules without also having to set integrations.collector explicitly (#2625) (@petewall)
• Add timeout setting on the OTLP destination (@petewall)
• Fail rendering with a clear error when profiling.enabled: true is set but none of profiling.ebpf.enabled, profiling.java.enabled, or profiling.pprof.enabled are enabled, instead of silently producing a profiling feature that collects no data (#2620) (@petewall)
• Update Alloy Operator to 0.5.8, OpenCost to 2.5.21 and Windows Exporter to 0.12.7 (@petewall)* Fix Alloy collector instances cross-contaminating each other's mounts.extra (and other nested values from collectorCommon.alloy or per-instance collectors.<name>) due to Sprig mergeOverwrite aliasing nested map/slice references. The user-supplied merge inputs ($userCommonValues and $userValues) are now deepCopy-ed before merging (#2623) (@petewall)

Full changelog: CHANGELOG.md