This release provides first-class support for running Prometheus in agent mode with the new PrometheusAgent CRD. As the v1alpha1 version tells it, we don't recommend using it in production but we're eager to hear all possible feedback.

To try it, follow the following steps:

1. Install the new CRD in the cluster (see example/prometheus-operator-crd/monitoring.coreos.com_prometheusagents.yaml).
2. Update the Prometheus operator's RBAC permissions to manage PrometheusAgents resources (see example/rbac/prometheus-operator/prometheus-operator-cluster-role.yaml).

NOTE: if these conditions aren't met, the operator will start but it won't be able to reconcile the PrometheusAgent resources.

For the first time, the container images associated to this release are signed using sigstore.

• [CHANGE] Remove the /apis endpoints from the operator's web server. #5396
• [CHANGE] Set default default value of spec.portName to web. #5350
• [FEATURE] Add v1alpha1 PrometheusAgent CRD to run Prometheus in agent mode. #5385
• [FEATURE] Add --reload-timeout argument to the config-reloader binary which controls how long the program will wait for the reload operation to complete (default: 30s). #5349
• [ENHANCEMENT] Set web server's ReadTimeout and ReadHeaderTimeout to 30s for Prometheus operator and config-reloader to avoid potential slowloris attacks. #5340
• [ENHANCEMENT] Add support for DropEqual and KeepEqual relabeling actions. #5368
• [ENHANCEMENT] Drop invalid PrometheusRule objects instead of failing the reconciliation of Prometheus and ThanosRuler objects. #5221
• [ENHANCEMENT] Add spec.thanos.blockSize field to the Prometheus CRD. #5360
• [ENHANCEMENT] Add spec.thanos.configTimeout and spec.thanos.configInterval to the Prometheus CRD. #5399
• [ENHANCEMENT] Add spec.alertmanagerConfiguration.global.slackApiUrl field to the Alertmanager CRD. #5383
• [ENHANCEMENT] Add spec.alertmanagerConfiguration.global.opsGenieApiUrl and spec.alertmanagerConfiguration.global.opsGenieApiKey fields to the Alertmanager CRD. #5422
• [ENHANCEMENT] Reduce the operator's memory usage by using metadata informers for Kubernetes secrets and configmaps. #5424 #5448
• [BUGFIX] Add init-config-reloader init container to avoid a restart of the Alertmanager's config-reloader container when the pod starts. #5358