TiDB Operator 1.4.0-alpha.1 Release Notes
Release date: September 27, 2022
TiDB Operator version: 1.4.0-alpha.1
Compatibility Change
- Due to changes in #4683, the volume modification feature is change to be disabled by default. If you want to resize PVCs of a component, you need to manually enable this feature.
Rolling Update Changes
- Due to changes in #4494, if you deploy TiCDC without setting the
log-file
configuration item, upgrading TiDB Operator to v1.4.0-alpha.1 causes TiCDC to rolling upgrade.
New Feature
Support setting location labels for tidb-server automatically (#4663, @glorv)
Add new fields
spec.tikv.scalePolicy
andspec.tiflash.scalePolicy
to scale multiple TiFlash and TiKV Pods simultaneously (#3881, @lizhemingi)Add a new field
startScriptVersion
to choose start script for all components (#4505, @KanShiori)Allow to use shortened label names to refer to some well-known Kubernetes labels in PD's location labels (#4688, @glorv)
Support point-in-time recovery from a snapshot backup and a log backup (#4648 #4682 #4694 #4695, @WizardXiao)
Add a new feature gate
VolumeModifying
to enable the volume modification feature, and the feature is disable by default (#4683, @liubog2008)Support changing the
StorageClass
of aTidbCluster
component to modify the IOPS and throughput for AWS EBS volumes used by a cluster (#4683, @liubog2008)Support setting the
--check-requirements
argument for BR (#4631, @KanShiori)Support using the field
additionalContainers
to customize pod container configuration. If the container names in this field match with the ones generated by TiDB Operator, the container configurations will be merged into the default configuration. (#4530, @mikechengwei)
Improvements
Optimize prometheus remoteWrite configuration for
TidbMonitor
(#4247, @mikechengwei)Add metrics port for TiFlash
Service
(#4470, @mikechengwei)Update the default value of the
log-file
configuration item for TiCDC to avoid overwriting/dev/stderr
(#4494, @KanShiori)
Bug fixes
Fix the issue that cluster sync gets stuck caused by suspending a cluster when scaling (#4679, @KanShiori)
Fix the issue that TiDB Operator panics if PD spec is nil (#4679, @KanShiori)