TiDB Operator 1.1.5 Release Notes

Release date: September 18, 2020

TiDB Operator version: 1.1.5

Compatibility Changes

• If the TiFlash version is earlier than v4.0.5, please set spec.tiflash.config.config.flash.service_addr: {clusterName}-tiflash-POD_NUM.{clusterName}-tiflash-peer.{namespace}.svc:3930 in the TidbCluster CR ({clusterName} and {namespace} need to be replaced with the real value) before upgrading TiDB Operator to v1.1.5 or later versions. When TiFlash is going to be upgraded to v4.0.5 or later versions, please remove spec.tiflash.config.config.flash.service_addr in the TidbCluster CR at the same time (#3191, @DanielZhangQD)

New Features

• Support configuring serviceAccount for TiDB/Pump/PD (#3246, @july2993)
• Support configuring spec.tikv.config.log-format and spec.tikv.config.server.max-grpc-send-msg-len (#3199, @kolbe)
• Support labels configuration for TiDB (#3188, @howardlau1999)
• Support recovery from failover for TiFlash and TiKV (#3189, @DanielZhangQD)
• Add mountClusterClientSecret configuration for PD and TiKV. If the config is set to true, TiDB Operator will mount the ${cluster_name}-cluster-client-secret to the PD or TiKV containers (#3282, @DanielZhangQD)

Improvements

• Adapt TiDB/PD/TiKV configurations to v4.0.6 (#3180, @lichunzhu)
• Support mounting the cluster client certificate to PD pod (#3248, @weekface)
• Scaling takes precedence over upgrading for TiFlash/PD/TiDB. This is to avoid that Pods cannot be scaled in case of upgrade failure (#3187, @lichunzhu)
• Support the imagePullSecrets configuration for Pump (#3214, @weekface)
• Update the default configuration for TiFlash (#3191, @DanielZhangQD)
• Remove ClusterRole from TidbMonitor CR (#3190, @weekface)
• Drainer that are deployed by Helm and exits normally will no longer be restarted (#3151, @lichunzhu)
• The tidb-scheduler HA strategy takes failover pods into consideration (#3171, @cofyc)

Bug Fixes

• Fix the problem that the Env settings are ignored for the Grafana container in the TidbMonitor CR (#3237, @tirsen)