TiDB Operator 1.1.10 Release Notes

Release date: January 28, 2021

TiDB Operator version: 1.1.10

Compatibility Changes

Due to the changes of #3638, the apiVersion of ClusterRoleBinding, ClusterRole, RoleBinding, and Role created in the TiDB Operator chart is changed from rbac.authorization .k8s.io/v1beta1 to rbac.authorization.k8s.io/v1. In this case, upgrading TiDB Operator through helm upgrade may report the following error:

Error: UPGRADE FAILED: rendered manifests contain a new resource that already exists. Unable to continue with update: existing resource conflict: namespace:, name: tidb-operator:tidb-controller-manager, existing_kind: rbac.authorization.k8s.io/ v1, Kind=ClusterRole, new_kind: rbac.authorization.k8s.io/v1, Kind=ClusterRole

For details, refer to helm/helm#7697. In this case, you need to delete TiDB Operator through helm uninstall and then reinstall it (deleting TiDB Operator will not affect the current TiDB clusters).

Rolling Update Changes

Upgrading TiDB Operator will cause the recreation of the TidbMonitor Pod due to #3684

New Features

Support canary upgrade of TiDB Operator (#3548, @shonge, #3554, @cvvz)
TidbMonitor supports remotewrite configuration (#3679, @mikechengwei)
Support configuring init containers for components in the TiDB cluster (#3713, @handlerww)
Add local backend support to the TiDB Lightning chart (#3644, @csuzhangxc)

Improvements

Support customizing the storage config for TiDB slow log (#3731, @BinChenn)
Add the tidb_cluster label for the scrape jobs in TidbMonitor to support monitoring multiple clusters (#3750, @mikechengwei)
Supports persisting checkpoint for the TiDB Lightning helm chart (#3653, @csuzhangxc)
Change the directory of the customized alert rules in TidbMonitor from tidb:${tidb_image_version} to tidb:${initializer_image_version} so that when the TiDB cluster is upgraded afterwards, the TidbMonitor Pod will not be recreated (#3684, @BinChenn)

Bug Fixes

Fix the issue that when TLS is enabled for the TiDB cluster, if spec.from or spec.to is not configured, backup and restore jobs with BR might fail (#3707, @BinChenn)
Fix the bug that if the advanced StatefulSet is enabled and delete-slots annotations are added for PD or TiKV, the Pods whose ordinal is bigger than replicas - 1 will be terminated directly without any pre-delete operations such as evicting leaders (#3702, @cvvz)
Fix the issue that after the Pod has been evicted or killed, the status of backup or restore is not updated to Failed (#3696, @csuzhangxc)
Fix the issue that when the TiKV cluster is not bootstrapped due to incorrect configuration, the TiKV component could not be recovered by editing TidbCluster CR (#3694, @cvvz)