TiDB Operator 1.1.7 Release Notes

Release date: November 13, 2020

TiDB Operator version: 1.1.7

Compatibility Changes

• The behavior of prometheus.spec.config.commandOptions has changed. Any duplicated flags must be removed, or Prometheus will fail to start. (#3390, @mightyguava)

  Flags that CANNOT be set are:

  • --web.enable-admin-api
  • --web.enable-lifecycle
  • --config.file
  • --storage.tsdb.path
  • --storage.tsdb.retention

New Features

• Support spec.toolImage for the Backup and Restore CR with BR to define the image used to provide the BR binary executables. Defaults to pingcap/br:${tikv_version} (#3471, @namco1992)
• Add spec.tidb.storageVolumes, spec.tikv.storageVolumes, and spec.pd.storageVolumes to support mounting multiple PVs for TiDB, TiKV, and PD (#3425 #3444, @mikechengwei)
• Add spec.tidb.readinessProbe config to support requesting http://127.0.0.0:10080/status for TiDB's readiness probe, TiDB version >= v4.0.9 required (#3438, @july2993)
• Support PD leader transfer with advanced StatefulSet controller enabled (#3395, @tangwz)
• Support setting OnDelete update strategies for the StatefulSets via spec.statefulSetUpdateStrategy in the TidbCluster CR (#3408, @cvvz)
• Support HA scheduling when failover happens (#3419, @cvvz)
• Support smooth migration from TiDB clusters deployed using TiDB Ansible or TiUP or deployed in the same Kubernetes cluster to a new TiDB cluster (#3226, @cvvz)
• tidb-scheduler supports advanced StatefulSet (#3388, @cvvz)

Improvements

• Forbid to scale in TiKV when the number of UP stores is equal to or less than 3 (#3367, @cvvz)
• phase is added in BackupStatus and RestoreStatus, which will be in sync with the latest condition type and shown when doing kubectl get (#3397, @namco1992)
• Skip setting tikv_gc_life_time via SQL for backup and restore with BR when the TiKV version >= v4.0.8 (#3443, @namco1992)

Bug Fixes

• Fix the issue that PD cannot scale in to zero if there are other PD members outside of this TidbCluster (#3456, @dragonly)