Releases: FoundationDB/fdb-kubernetes-operator
Releases · FoundationDB/fdb-kubernetes-operator
Release v0.35.1
v0.35.1
- Prevent sub-optimal recruitment of coordinators in multi-DC configurations.
- Remove use of recently deprecated paths in initializing the controllers.
Release v0.35.0
v0.35.0
- Fix a bug with the arguments to the backup and restore commands.
- Refactor the setup in the main method to move more setup code into a library.
- Merge the steps for confirming exclusion completion and removing the pods.
- Improve handling of missing pods when updating the status.
- Change our release process to upload the binary for the plugin instead of an
archive. - Disable end-to-end tests by default when running
go test
directly. - Use different config map hashes for different process classes.
- Remove unnecessary requeueing of reconciliation when a step updates the
status. - Move the FDB client dependencies into a dedicated subpackage.
- Prefer to place coordinators on storage pods.
- Add the metadata from the pod template to the schema to prevent pruning.
- Add an option in the spec to prevent reconciling a cluster.
Release v0.34.1
patch version bump
Release v0.34.0
version bump
Release v0.33.0
v0.33.0
- Ignore processes that are pending removal when doing a bounce.
- Simplify selection of FDB version when generating the sidecar image version.
- Improve support for custom sidecar images.
- Add initial end-to-end tests.
- Add a restart command to the kubectl plugin.
- Add an analyze command to the kubectl plugin.
- Validate FDB version format as part of the CRD definition.
Release v0.32.0
version bump to v0.32.0
Release v0.31.1
v0.31.1
- Remove the custom CA certs used when downloading artifacts from the FoundationDB website.
Release v0.31.0
v0.31.0
- Block exclusions until replacement processes are reporting to the cluster.
- Prevent reconciliation from getting stuck when pods are stuck in terminating
and have failing conditions on the process group. - Add the age of the FDB object to the kubectl output.
- Improve parallelism when checking for updates to config maps.
- Update the version of the CRD API we use in our CRD definition.
- Update logging flags to conform with standard names for Zap.
- Fix a bug that caused error events to be silently lost.
- Requeue reconciliation when we detect an increase to the resource version for
the cluster resource. - Use patch requests when making changes to the cluster spec in the kubectl
plugin. - Fix Prometheus configuration in the sample deployment.
- Allow specifying a custom duration for the minimum uptime between bounces.
Release v0.30.0
v0.30.0
- Add tini to the sidecar image.
- Use locality information to determine which processes to remove in a shrink.
- Prevent reconciliation from being blocked when pods are stuck in terminating.
- Add documentation on working with locks.
- Test the CRDs against multiple versions of the API server.
- Manage more conditions through the process group status.
- Represent process classes as a customn type.
- Fix the image tag in the helm chart.
- Add validation on the satellite configuration in the region config.
Release v0.29.0
v0.29.0
- Fix #520: Public NotReady pods to pod cluster IPs.
- Add test for kubectl rootcmd.
- Prevent removing unknown labels in pods.
- Allow removing instances by pod name.
- Update controller-runtime and dependencies.
- Add deny list to prevent specific instances of the operator from taking global operations.
- Rename release job.
- Add locks before removing pods or including instances.
- Add python lint.
- Update to golang 1.15.8.
- Add evacuate command for plugin.