// the find
kubernetes-monitoring/kubernetes-mixin
A set of Grafana dashboards and Prometheus alerts for Kubernetes.
A collection of Jsonnet-templated Grafana dashboards and Prometheus alert rules covering the full Kubernetes control plane and workload stack. Designed to be vendored into your own monitoring config and customized via Jsonnet mixins. This is the upstream source for what kube-prometheus and many Helm-based stacks ship out of the box.
The mixin pattern is genuinely useful — you override `_config` in Jsonnet rather than forking YAML files, so you can pull upstream updates without losing your customizations. Alert severity tiers (critical/warning/info) are defined with actual operational intent, not just 'this fired'. The compatibility matrix is maintained and honest about which Kubernetes and kube-state-metrics versions are supported. Windows node dashboards are included, which most Kubernetes monitoring projects still treat as an afterthought.
Jsonnet is a high barrier to entry — most platform engineers know YAML and PromQL, not a data templating language with a non-obvious import/vendoring model. The ksonnet integration in the README is dead (ksonnet was abandoned in 2019) and the prometheus-operator section is still 'TODO', which is ironic given that's how most people actually run this today. The dev setup spins up a kind cluster, which is solid, but the feedback loop for alert changes requires a manual `make dev-reload` step with no hot-reload. Alert tests in `tests/` are PromQL unit tests, not end-to-end — you can have a passing test suite and still have alerts that never fire in production due to label mismatches.