Helm operator 0.10.0

This release brings you opt-in automated rollback support, new Prometheus metrics, and experimental support of spawning multiple workers with the --workers=<num> flag to speed up the processing of releases.

This will likely also be the last minor beta release before we promote the Helm operator to its first GA 1.0.0 release.

Notice: the Helm operator relies on changes in the HelmRelease Custom Resource Definition. Please make sure you patch the CRD in your cluster before upgrading the Helm operator.

Bug fixes

• Prevent an infinite release loop when multiple HelmRelease resources with the same release name configuration coexist, by looking at the antecedent annotation set on release resources and confirming ownership weaveworks/flux#2123

Improvements

• Opt-in automated rollback support; when enabled, a failed release will be rolled back automatically and the operator will not attempt a new release until it detects a change in the chart and/or the configured values weaveworks/flux#2006

• Increase timeout for annotating resources from a Helm release, to cope with large umbrella charts weaveworks/flux#2123

• New Prometheus metrics

  • release_queue_length_count
  • release_duration_seconds{action=['INSTALL','UPGRADE'], dry-run=['true', 'false'], success=['true','false'], namespace, releasename}

  weaveworks/flux#2191

• Experimental support of spawning multiple queue workers processing releases by configuring the --workers=<num> flag weaveworks/flux#2194

Maintenance and documentation

• Publish images to fluxcd DockerHub organization weaveworks/flux#2213
• Document opt-in rollback feature weaveworks/flux#2220

Thanks

Many thanks to @adrian, @2opremio, @semyonslepov, @gtseres, @squaremo, @stefanprodan, @kingdonb, @ncabatoff, @dholbach, @cristian-radu, @simonmacklin, @hiddeco for contributing to this release.