TiDB Operator 1.5.0 Release Notes

Release date: August 4, 2023

TiDB Operator version: 1.5.0

Rolling update changes

If TiFlash is deployed in a TiDB cluster that is v7.1.0 or later, the TiFlash component will be rolling updated after TiDB Operator is upgraded to v1.5.0 due to #5075.

New features

• Add the BR Federation Manager component to orchestrate Backup and Restore custom resources (CR) across multiple Kubernetes clusters (#4996, @csuzhangxc)
• Support using the VolumeBackup CR to back up a TiDB cluster deployed across multiple Kubernetes clusters based on EBS snapshots (#5013, @WangLe1321)
• Support using the VolumeRestore CR to restore a TiDB cluster deployed across multiple Kubernetes clusters based on EBS snapshots (#5039, @WangLe1321)
• Support using the VolumeBackupSchedule CR to automatically back up a TiDB cluster deployed across multiple Kubernetes clusters based on EBS snapshots (#5036, @BornChanger)
• Support backing up CRs related to TidbCluster when backing up a TiDB cluster deployed across multiple Kubernetes based on EBS snapshots (#5207, @WangLe1321)

Improvements

• Add the startUpScriptVersion field for DM master to specify the version of the startup script (#4971, @hanlins)
• Support spec.preferIPv6 for DmCluster, TidbDashboard, TidbMonitor, and TidbNGMonitoring (#4977, @KanShiori)
• Support setting expiration time for TiKV leader eviction and PD leader transfer (#4997, @Tema)
• Support setting toleration for TidbInitializer (#5047, @csuzhangxc)
• Support configuring the timeout for PD start (#5071, @oliviachenairbnb)
• Skip evicting leaders for TiKV when changing PVC size to avoid leader eviction blocked caused by low disk space (#5101, @csuzhangxc)
• Support updating annotations and labels in services for PD, TiKV, TiFlash, TiProxy, DM-master, and DM-worker (#4973, @wxiaomou)
• Enable volume resizing by default for PV expansion (#5167, @liubog2008)

Bug fixes

• Fix the quorum loss issue during TiKV upgrade due to some TiKV stores going down (#4979, @Tema)
• Fix the quorum loss issue during PD upgrade due to some members going down (#4995, @Tema)
• Fix the issue that TiDB Operator panics when no Kubernetes cluster-level permission is configured (#5058, @liubog2008)
• Fix the issue that TiDB Operator might panic when AdditionalVolumeMounts is set for the TidbCluster CR (#5058, @liubog2008)
• Fix the issue that baseImage for the TidbDashboard CR is parsed incorrectly when custom image registry is used (#5014, @linkinghack)