Releases: FoundationDB/fdb-kubernetes-operator
Releases · FoundationDB/fdb-kubernetes-operator
v2.4.0
v2.3.0
v2.3.0
Changes
Operator
- Ensure that the cordon command doesn't stop when a node doesn't host any pods #2260
- Implement changes for Multi-Cluster coordination between multiple operator instances. #2259
- Add a dedicated condition for wrong pod metadata #2258
- Remove the last usage of the pvc map method #2257
- Fix test case for Pod IP family change #2256
- Add support for using patch to update pods instead of update #2253
v2.2.0
v2.2.0
Changes
Operator
- Update golang and xnet #2251
- Ensure that pods are replaced if we change the Pod IP family #2247
- Ensure that the operator checks if the processes for newly created pods are up and running for the update pod config reconciler #2244
- Fix a bug in the connection string parser when a cluster has a single letter as name #2240
- Correct check for incompatible clients if the coordinator pod hosted more than one process #2238
- Changes in update database configuration to detect if the database was configured #2237
- Check the current cluster status before removing maintenance mode #2235
- Improvements to e2e tests and documentation #2233
- Change the ordering of the coordinator change and the exclusion #2232
- Update cluster file usage. #2230
v2.1.0
v2.1.0
Changes
Operator
- Allow to modify the default CPU and memory for pods #2236
- Remove the 3DH setup from e2e pipeline to reduce required resources #2231
- Avoid panics when merging custom annotation/labels #2227
- Fix PR action for docker build #2225
- Update the e2e tests to reflect the new defaults #2222
- Don't run backup tests for bad version #2221
- Add a new label for running a quick upgrade test case #2201
v2.0.0
v2.0.0
Changes
Action Required
Please read the required steps before upgrading in the compatibility guide.
Breaking Changes
- Dropped support for the
v1beta1CRD #2212 - Use locality-based exclusions per default #2212
- Use DNS in cluster files per default #2212
- Use the unified image per default #2212
- Drop support for FDB 6.2 and 6.3 #2212
- Drop support for custom PVC names #2213
- Drop support for custom ConfigMap names #2215
Operator
v1.55.0
v1.55.0
Changes
Operator
- Ensure that the tag suffix is only added when it is missing #2208
- Check if the newly added process group ID is already excluded #2207
- Update docs for three data hall and move something to implemented #2206
- Ensure reconcilation is requeued for restore controller if the restore is not completed #2203
- Allow to modify the e2e peer verification string #2202
- Add the restore state to the restore CRD #2198
- Mark the HA upgrade test where no remote processes are restarted as pending until it passes reliably #2197
- Ensure the init container is removed if the unified image should be used #2195
- docs: document procedure to migrate a live cluster to three-data-hall redundancy #2191
v1.54.0
v1.53.0
v1.53.0
Changes
Operator
- Update the x/net dependency. #2189
- Update the three_data_hall docs for the three data hall with unified image #2188
- Fix the plugin e2e test setup #2184
- Add a test case to run the exclusion in a multi-region cluster for a bigger cluster with injected latencies #2183
- Update go version to 1.22.10 and RockyLinux to 9.5 #2182
- Add support for multi-region recover subcommand with DNS #2181
- Correct the failure duration for unavailable checks #2178
- Correct init container setup for unified image with backups #2177
- Improve the reliability of the pod pending test case #2176
- Always make use of the no_wait flag in the exclude controller to reduce the noise from exclusions that timeout #2175
- Enable the unified image randomly in the e2e tests #2173
- Add design for better coordination for multi operator setup #2172
- Reset the exclusion status if a process is not excluded anymore #2167
v1.52.0
v1.51.0
v1.51.0
Changes
Operator
- Try to batch the exclusions of transaction system processes together #2158
- Smaller doc fixes #2157
- Fix the unified image setup during bootstrap when a custom environment variable is used for the zone #2156
- Make sure that the remove cluster is skipped too and add more debugging output for the plugin #2154
- Update go version to 1.22.8 #2152
- Simplify the confirmRemoval code #2148