Warning
This release is unstable and should not be used in production.
Helm chart versions <0.20.0 were built using manifests from the main branch at release time instead of the release branch. As a result, deployed manifests may not match the expected code version.
Required action:
Upgrade to v1.3.1 (Helm Chart 0.20.0). This is the first version with corrected manifests and a tested upgrade path from affected releases.
Recommended next step:
After upgrading to v1.3.1, continue upgrading to the latest GA release (v1.7.0) for production use.
For production environments, upgrade one minor version at a time after 0.20.0.
See docs/release.md for full details on versioning and upgrade paths.
Warning
This release is unstable and should not be used in production.
The kube-rbac-proxy project has officially announced the deprecation of gcr.io/kubebuilder/kube-rbac-proxy in favor of registry.k8s.io. The controller Helm chart in this release and in earlier releases references the now-unavailable gcr.io/kubebuilder/kube-rbac-proxy image. The controller will therefore fail to start up with ImagePullBackoff.
The issue is fixed in #219 and #222, so any subsequent releases with either of those PRs will solve this issue.
Recommended action: Skip this version and upgrade to a release containing the fix. We are running validations on v1.1.2 now and will mark it as latest ASAP.
Highlights
PR #163 fixes a bug that arises when v1.1.0 of the controller tries to roll out the version after matching.maxVersionsInDeployment (so the 101st version by default). This will also be fixed in v1.29.2 of the open source server, so users can avoid hitting the bug either by upgrading their server or by upgrading to v1.1.1 of the controller.
If you do hit this bug and cannot upgrade your controller or server version right away, see docs/ownership.md for instructions to move forward with your deployment.
The symptom of the bug will be that your 101st version is stuck in "Inactive" state, and the controller will log error messages pointing to docs/ownership.md. The controller needs to be prompted via the Temporal CLI to move forward with the rollout.
What's Changed
- feat: add support for gating rollouts behind successful workflow executions by @kianjones9 in #152
- Refactor integration tests so they can be run one at a time in IDE by @carlydf in #155
- Document API key setup and add details about secret creation by @carlydf in #160
- Ownership docs: Update the docs to reflect the right command. by @Shivs11 in #161
- VLN-516: Set explicit permissions for GitHub Actions workflows by @picatz in #159
- Bug Fix: Ignore LastModifierIdentity if server deleted a version for garbage collection by @carlydf in #163
New Contributors
- @kianjones9 made their first contribution in #152
- @picatz made their first contribution in #159
Full Changelog: v1.1.0...v1.1.1