TiDB Operator 1.3.0-beta.1 Release Notes
TiDB Operator version: 1.3.0-beta.1
Due to changes in #4151, if you deploy v1 CRD, TiDB Operator >= v1.3.0-beta.1 sets the default
baseImage
field of all components. If you set the component image using the field instead of thebaseImage
field, upgrading TiDB Operator to v1.3.0-beta.1 will change the image in use, cause the TiDB cluster to rolling update or even fail to run. To avoid such situations, you must upgrade TiDB Operator by the following steps:- Use the
baseImage
andversion
fields to replace theimage
field. For details, refer to . - Upgrade TiDB Operator.
- Use the
- Due to changes in #4358, if the TiDB cluster (>= v5.4) is deployed by TiDB Operator v1.2, upgrading TiDB Operator to v1.3.0-beta.1 causes TiFlash to rolling upgrade. It is recommended to upgrade TiDB Operator to v1.3 before upgrading the TiDB cluster to v5.4.0 or later versions.
- Due to changes in , for TiDB clusters >= v5.0, if
spec.tikv.seperateRocksDBLog: true
orspec.tikv.separateRaftLog: true
is configured, upgrading TiDB Operator to v1.3.0-beta.1 causes TiKV to rolling upgrade. - Due to changes in #4198, upgrading TiDB Operator causes the recreation of TidbMonitor Pod.
- Support configuring the resource usage for the init container of TiFlash (, @KanShiori)
- Support enabling for the TiDB cluster (#4287, )
- Support gracefully restarting TiKV through annotations (#4279, )
- Support
PodSecurityContext
and other configurations for Discovery (#4259, , #4208, ) - Support configuring
PodManagementPolicy
in TidbCluster CR (#4211, ) - Support configuring Prometheus shards in TidbMonitor CR (#4198, )
- Generate v1 CRD to support deploying in Kubernetes v1.22 or later versions (#4151, )
- Remove and change some default configurations for TiFlash due to configuration changes in TiFlash v5.4.0. If the TiDB cluster (>= v5.4) is deployed by TiDB Operator v1.2, upgrading TiDB Operator to v1.3.0-beta.1 causes TiFlash to rolling upgrade. (#4358, )
- Improve advanced deployment example of TidbMonitor. (#4242, )
- Optimize the user experience of heterogenous clusters by displaying the metrics for one TiDB cluster and its heterogeneous clusters in the same dashboards. (#4210, )
- Use to obtain Grafana password in TidbMonitor to avoid using plaintext password. (#4135, )
- Optimize the upgrade process for PD and TiKV components with fewer than two replicas, and force the upgrade of PD and TiKV components by default to avoid the upgrade process from taking too long (#4107, , @mianhk)
- Update Grafana images in examples to 7.5.11 to enhance security (, @makocchi-git)
- Deprecate Pod validating and mutating webhook (, @mianhk)
- Support configuring the number of tidb-controller-manager workers in Helm chart (, @mianhk)