Skip to content

Latest commit

 

History

History
435 lines (355 loc) · 57.1 KB

CHANGELOG-1.1.md

File metadata and controls

435 lines (355 loc) · 57.1 KB
Raw

Note:

  • Starting from September 2, 2020, all release notes of TiDB Operator will be maintained in pingcap/docs-tidb-operator.
  • You can read the release notes of all versions of TiDB Operator at PingCAP Docs.

TiDB Operator v1.1.4 Release Notes

Notable changes

  • TableFilter is added to the BackupSpec and RestoreSpec. TableFilter supports backing up specific databases or tables with Dumpling or BR and supports restoring specific databases or tables with BR. BackupSpec.Dumpling.TableFilter is deprecated since v1.1.4. Please configure BackupSpec.TableFilter instead. Since TiDB v4.0.3, you can configure BackupSpec.TableFilter to replace the BackupSpec.BR.DB and BackupSpec.BR.Table fields and configure RestoreSpec.TableFilter to replace the RestoreSpec.BR.DB and RestoreSpec.BR.Table fields (#3134, @sstubbs)
  • Update the version of TiDB and tools to v4.0.4 (#3135, @lichunzhu)
  • Support customizing environment variables for the initializer container in the TidbMonitor CR (#3109, @kolbe)
  • Support patching PVCs when the storage request is increased (#3096, @cofyc)
  • Support TLS for Backup & Restore with Dumpling & TiDB Lightning (#3100, @lichunzhu)
  • Support cert-allowed-cn for TiFlash (#3101, @DanielZhangQD)
  • Add support for the max-index-length TiDB config option to the TidbCluster CRD (#3076, @kolbe)
  • Fix goroutine leak when TLS is enabled (#3081, @DanielZhangQD)
  • Fix a memory leak issue caused by etcd client when TLS is enabled (#3064, @DanielZhangQD)
  • Support TLS for TiFlash (#3049, @DanielZhangQD)
  • Configure TZ environment for admission webhook and advanced statefulset controller deployed in tidb-operator chart (#3034, @cofyc)

TiDB Operator v1.1.3 Release Notes

Action Required

  • Add a field cleanPolicy in BackupSpec to denote the clean policy for backup data when the Backup CR is deleted from the cluster (default to Retain). Note that before v1.1.3, TiDB Operator will clean the backup data in the remote storage when the Backup CR is deleted, so if you want to clean backup data as before, set spec.cleanPolicy in Backup CR or spec.backupTemplate.cleanPolicy in BackupSchedule CR to Delete. (#3002, @lichunzhu)
  • Replace mydumper with dumpling for backup. If spec.mydumper is configured in the Backup CR or spec.backupTemplate.mydumper is configured in the BackupSchedule CR, migrate it to spec.dumpling or spec.backupTemplate.dumpling. After you upgrade TiDB Operator to v1.1.3, note that spec.mydumper or spec.backupTemplate.mydumper will be lost after the upgrade. (#2870, @lichunzhu)

Other Notable Changes

  • Update tools in backup manager to v4.0.3 (#3019, @lichunzhu)
  • Support cleanPolicy for the Backup CR to define the clean behavior of the backup data in the remote storage when the Backup CR is deleted (#3002, @lichunzhu)
  • Add TLS support for TiCDC (#3011, @weekface)
  • Add TLS support between Drainer and the downstream database server (#2993, @lichunzhu)
  • Support specifying mysqlNodePort and statusNodePort for TiDB Service Spec (#2941, @lichunzhu)
  • Fix the initialCommitTs bug in Drainer's values.yaml (#2857, @weekface)
  • Add backup config for TiKV server, add enable-telemetry, and deprecate disable-telemetry config for PD server (#2964, @lichunzhu)
  • Add commitTS info column in get restore command (#2926, @lichunzhu)
  • Update the used Grafana version from v6.0.1 to v6.1.6 (#2923, @lichunzhu)
  • Support showing commitTS in restore status (#2899, @lichunzhu)
  • Exit without error if the backup data the user tries to clean does not exist (#2916, @lichunzhu)
  • Support auto-scaling by storage for TiKV in TidbClusterAutoScaler (#2884, @Yisaer)
  • Clean temporary files in Backup job with Dumpling to save space (#2897, @lichunzhu)
  • Fail the backup job if existing PVC's size is smaller than the storage request in the backup job (#2894, @lichunzhu)
  • Support scaling and auto-failover even if a TiKV store fails in upgrading (#2886, @cofyc)
  • Fix a bug that the TidbMonitor resource could not be set (#2878, @weekface)
  • Fix an error for the monitor creation in the tidb-cluster chart (#2869, @8398a7)
  • Remove readyToScaleThresholdSeconds in TidbClusterAutoScaler; TiDB Operator won't support de-noise in TidbClusterAutoScaler (#2862, @Yisaer)
  • Update the version of TiDB Lightning used in tidb-backup-manager from v3.0.15 to v4.0.2 (#2865, @lichunzhu)

TiDB Operator v1.1.2 Release Notes

Action Required

  • An incompatible issue with PD 4.0.2 has been fixed. Please upgrade TiDB Operator to v1.1.2 before deploying TiDB 4.0.2 and later versions (#2809, @cofyc)

Other Notable Changes

  • Collect metrics for TiCDC, TiDB Lightning and TiKV Importer (#2835, @weekface)
  • Update PD/TiDB/TiKV config to v4.0.2 (#2828, @DanielZhangQD)
  • Fix the bug that PD Member might still exist after scaling-in (#2793, @Yisaer)
  • Support Auto-Scaler Reference in TidbCluster Status when there exists TidbClusterAutoScaler (#2791, @Yisaer)
  • Support configuring container lifecycle hooks and terminationGracePeriodSeconds in TiDB spec (#2810, @weekface)

TiDB Operator v1.1.1 Release Notes

Notable changes

  • Add the additionalContainers and additionalVolumes fields so that TiDB Operator can support adding sidecars to TiDB, TiKV, PD, etc. (#2229, @yeya24)
  • Add cross check to ensure TiKV is not scaled or upgraded at the same time (#2705, @DanielZhangQD)
  • Fix the bug that TidbMonitor will scrape multi TidbCluster with the same name in different namespaces when then namespace in ClusterRef is not set (#2746, @Yisaer)
  • Update TiDB Operator examples to deploy TiDB Cluster 4.0.0 images (#2600, @kolbe)
  • Add the alertMangerAlertVersion option to TidbMonitor (#2744, @weekface)
  • Fix alert rules lost after rolling upgrade (#2715, @weekface)
  • Fix an issue that pods may be stuck in pending for a long time in scale-out after a scale-in (#2709, @cofyc)
  • Add EnableDashboardInternalProxy in PDSpec to let user directly visit PD Dashboard (#2713, @Yisaer)
  • Fix the PV syncing error when TidbMonitor and TidbCluster have different values in reclaimPolicy (#2707, @Yisaer)
  • Update Configuration to v4.0.1 (#2702, @Yisaer)
  • Change tidb-discovery strategy type to Recreate to fix the bug that more than one discovery pod may exist (#2701, @weekface)
  • Expose the Dashboard service with HTTP endpoint whether tlsCluster is enabled (#2684, @Yisaer)
  • Add the .tikv.dataSubDir field to specify subdirectory within the data volume to store TiKV data (#2682, @cofyc)
  • Add the imagePullSecrets attribute to all components (#2679, @weekface)
  • Enable StatefulSet and Pod validation webhook to work at the same time (#2664, @Yisaer)
  • Emit an event if it fails to sync labels to TiKV stores (#2587, @PengJi)
  • Make datasource information hidden in log for Backup and Restore jobs (#2652, @Yisaer)
  • Support the DynamicConfiguration switch in TidbCluster Spec (#2539, @Yisaer)
  • Support LoadBalancerSourceRanges in the ServiceSpec for the TidbCluster and TidbMonitor (#2610, @shonge)
  • Support Dashboard metrics ability for TidbCluster when TidbMonitor deployed (#2483, @Yisaer)
  • Bump the DM version to v2.0.0-beta.1 (#2615, @tennix)
  • support setting discovery resources (#2434, @shonge)
  • Support the Denoising for the TidbCluster Auto-scaling (#2307, @vincent178)
  • Support scraping Pump and Drainer metrics in TidbMonitor (#2750, @Yisaer)

TiDB Operator v1.1.0 Release Notes

This is the GA release of TiDB Operator 1.1, which focuses on the usability, extensibility and security.

See our official documentation site for new features, guides, and instructions in production, etc.

Upgrade from v1.0.x

For v1.0.x users, refer to Upgrade TiDB Operator to upgrade TiDB Operator in your cluster. Note that you should read the release notes (especially breaking changes and action required items) before the upgrade.

Breaking changes since v1.0.0

  • Change TiDB pod readiness probe from HTTPGet to TCPSocket 4000 port. This will trigger rolling-upgrade for the tidb-server component. You can set spec.paused to true before upgrading tidb-operator to avoid the rolling upgrade, and set it back to false when you are ready to upgrade your TiDB server (#2139, @weekface)

  • --advertise-address is configured for tidb-server, which would trigger rolling-upgrade for the TiDB server. You can set spec.paused to true before upgrading TiDB Operator to avoid the rolling upgrade, and set it back to false when you are ready to upgrade your TiDB server (#2076, @cofyc)

  • --default-storage-class-name and --default-backup-storage-class-name flags are abandoned, and the storage class defaults to Kubernetes default storage class right now. If you have set default storage class different than Kubernetes default storage class, set them explicitly in your TiDB cluster Helm or YAML files. (#1581, @cofyc)

  • Add the timezone support for all charts (#1122, @weekface).

    For the tidb-cluster chart, we already have the timezone option (UTC by default). If the user does not change it to a different value (for example, Asia/Shanghai), none of the Pods will be recreated. If the user changes it to another value (for example, Aisa/Shanghai), all the related Pods (add a TZ env) will be recreated, namely rolling updated.

    The related Pods include pump, drainer, discovery, monitor, scheduled backup, tidb-initializer, and tikv-importer.

    All images' time zone maintained by TiDB Operator is UTC. If you use your own images, you need to make sure that the time zone inside your images is UTC.

Other Notable changes

Previous releases

TiDB Operator v1.1.0-rc.4 Release Notes

This is the fourth release candidate of v1.1.0, which focuses on the usability, extensibility and security of TiDB Operator. While we encourage usage in non-critical environments, it is NOT recommended to use this version in critical environments.

Action Required

  • Separate TiDB client certificates can be used for each component. Users should migrate the old TLS configs of Backup and Restore to the new configs. Refer to #2403 for more details (#2403, @weekface)

Other Notable Changes

  • Fix the bug that the service annotations would be exposed in TidbCluster specification (#2471, @Yisaer)
  • Fix a bug when reconciling TiDB service while the healthCheckNodePort is already generated by Kubernetes (#2438, @aylei)
  • Support TidbMonitorRef in TidbCluster Status (#2424, @Yisaer)
  • Support setting the backup path prefix for remote storage (#2435, @onlymellb)
  • Support customizing mydumper options in Backup CR (#2407, @onlymellb)
  • Support TiCDC in TidbCluster CR. (#2338, @weekface)
  • Update BR to v3.1.1 in the tidb-backup-manager image (#2425, @DanielZhangQD)
  • Support creating node pools for TiFlash and CDC on ACK (#2420, @DanielZhangQD)
  • Support creating node pools for TiFlash and CDC on EKS (#2413, @DanielZhangQD)
  • Expose PVReclaimPolicy for TidbMonitor when storage is enabled (#2379, @Yisaer)
  • Support arbitrary topology-based HA in tidb-scheduler (e.g. node zones) (#2366, @PengJi)
  • Skip setting the TLS for PD dashboard when the TiDB version is earlier than 4.0.0 (#2389, @weekface)
  • Support backup and restore with GCS using BR (#2267, @shuijing198799)
  • Update TiDBConfig and TiKVConfig to support the 4.0.0-rc version (#2322, @Yisaer)
  • Fix the bug when TidbCluster service type is NodePort, the value of NodePort would change frequently (#2284, @Yisaer)
  • Add external strategy ability for TidbClusterAutoScaler (#2279, @Yisaer)
  • PVC will not be deleted when TidbMonitor gets deleted (#2374, @Yisaer)
  • Support scaling for TiFlash (#2237, @DanielZhangQD)

TiDB Operator v1.1.0-rc.3 Release Notes

This is the third release candidate of v1.1.0, which focuses on the usability, extensibility and security of TiDB Operator. While we encourage usage in non-critical environments, it is NOT recommended to use this version in critical environments.

Notable Changes

  • Skip auto-failover when pods are not scheduled and perform recovery operation no matter what state failover pods are in (#2263, @cofyc)
  • Support TiFlash metrics in TidbMonitor (#2341, @Yisaer)
  • Do not print rclone config in the Pod logs (#2343, @DanielZhangQD)
  • Using Patch in periodicity controller to avoid updating StatefulSet to the wrong state (#2332, @Yisaer)
  • Set enable-placement-rules to true for PD if TiFlash is enabled in the cluster (#2328, @DanielZhangQD)
  • Support rclone options in the Backup and Restore CR (#2318, @DanielZhangQD)
  • Fix the issue that statefulsets are updated during each sync even if no changes are made to the config (#2308, @DanielZhangQD)
  • Support configuring Ingress in TidbMonitor (#2314, @Yisaer)
  • Fix a bug that auto-created failover pods can't be deleted when they are in the failed state (#2300, @cofyc)
  • Add useful Event in TidbCluster during upgrading and scaling when admissionWebhook.validation.pods in operator configuration is enabled (#2305, @Yisaer)
  • Fix the issue that services are updated during each sync even if no changes are made to the service configuration (#2299, @DanielZhangQD)
  • Fix a bug that would cause panic in statefulset webhook when the update strategy of StatefulSet is not RollingUpdate (#2291, @Yisaer)
  • Fix a panic in syncing TidbClusterAutoScaler status when the target TidbCluster does not exist (#2289, @Yisaer)
  • Fix the pdapi cache issue while the cluster TLS is enabled (#2275, @weekface)
  • Fix the config error in restore (#2250, @Yisaer)
  • Support failover for TiFlash (#2249, @DanielZhangQD)
  • Update the default eks version in terraform scripts to 1.15 (#2238, @Yisaer)
  • Support upgrading for TiFlash (#2246, @DanielZhangQD)
  • Add stderr logs from BR to the backup-manager logs (#2213, @DanielZhangQD)
  • Add field TiKVEncryptionConfig in TiKVConfig, which defines how to encrypt data key and raw data in TiKV, and how to back up and restore the master key. See the description for details in tikv_config.go (#2151, @shuijing198799)

TiDB Operator v1.1.0-rc.2 Release Notes

This is the second release candidate of v1.1.0, which focuses on the usability, extensibility and security of TiDB Operator. While we encourage usage in non-critical environments, it is NOT recommended to use this version in critical environments.

Action Required

  • Change TiDB pod readiness probe from HTTPGet to TCPSocket 4000 port. This will trigger rolling-upgrade for the tidb-server component. You can set spec.paused to true before upgrading tidb-operator to avoid the rolling upgrade, and set it back to false when you are ready to upgrade your tidb server (#2139, @weekface)

Notable Changes

  • Add status field for TidbAutoScaler CR (#2182, @Yisaer)
  • Add spec.pd.maxFailoverCount field to limit max failover replicas for PD (#2184, @cofyc)
  • Emit more events for TidbCluster and TidbClusterAutoScaler to help users know TiDB running status (#2150, @Yisaer)
  • Add the AGE column to show creation timestamp for all CRDs (#2168, @cofyc)
  • Add a switch to skip PD Dashboard TLS configuration (#2143, @weekface)
  • Support deploying TiFlash with TidbCluster CR (#2157, @DanielZhangQD)
  • Add TLS support for TiKV metrics API (#2137, @weekface)
  • Set PD DashboardConfig when TLS between the MySQL client and TiDB server is enabled (#2085, @weekface)
  • Remove unnecessary informer caches to reduce the memory footprint of tidb-controller-manager (#1504, @aylei)
  • Fix the failure that Helm cannot load the kubeconfig file when deleting the tidb-operator release during terraform destroy (#2148, @DanielZhangQD)
  • Support configuring the Webhook TLS setting by loading a secret (#2135, @Yisaer)
  • Support TiFlash in TidbCluster CR (#2122, @DanielZhangQD)
  • Fix the error that alertmanager couldn't be set in TidbMonitor (#2108, @Yisaer)

TiDB Operator v1.1.0-rc.1 Release Notes

This is a release candidate of v1.1.0, which focuses on the usability, extensibility and security of TiDB Operator. While we encourage usage in non-critical environments, it is NOT recommended to use this version in critical environments.

Action Required

  • --advertise-address will be configured for tidb-server, which would trigger rolling-upgrade for the tidb-server component. You can set spec.paused to true before upgrading tidb-operator to avoid the rolling upgrade, and set it back to false when you are ready to upgrade your tidb server (#2076, @cofyc)
  • Add the tlsClient.tlsSecret field in the backup and restore spec, which supports specifying a secret name that includes the cert (#2003, @shuijing198799)
  • Remove spec.br.pd, spec.br.ca, spec.br.cert, spec.br.key and add spec.br.cluster, spec.br.clusterNamespace for the Backup, Restore and BackupSchedule custom resources, which makes the BR configuration more reasonable (#1836, @shuijing198799)

Other Notable Changes

  • Use tidb-lightning in Restore instead of loader (#2068, @Yisaer)
  • Add cert-allowed-cn support to TiDB components (#2061, @weekface)
  • Fix the PD location-labels configuration (#1941, @aylei)
  • Able to pause and unpause tidb cluster deployment via spec.paused (#2013, @cofyc)
  • Default the max-backups for TiDB server configuration to 3 if the TiDB cluster is deployed by CR (#2045, @Yisaer)
  • Able to configure custom environments for components (#2052, @cofyc)
  • Fix the error that kubectl get tc cannot show correct images (#2031, @Yisaer)
    1. Default the spec.tikv.maxFailoverCount and spec.tidb.maxFailoverCount to 3 when they are not defined
    2. Disable auto-failover when maxFailoverCount is set to 0 (#2015, @Yisaer)
  • Support deploying TiDB clusters with TidbCluster and TidbMonitor CRs via Terraform on ACK (#2012, @DanielZhangQD)
  • Update PDConfig for TidbCluster to PD v3.1.0 (#1928, @Yisaer)
  • Support deploying TiDB clusters with TidbCluster and TidbMonitor CRs via Terraform on AWS (#2004, @DanielZhangQD)
  • Update TidbConfig for TidbCluster to TiDB v3.1.0 (#1906, @Yisaer)
  • Allow users to define resources for initContainers in TiDB initializer job (#1938, @tfulcrand)
  • Add TLS support for Pump and Drainer (#1979, @weekface)
  • Add documents and examples for auto-scaler and initializer (#1772, @Yisaer)
    1. Add check to guarantee the NodePort won't be changed if the serviceType of TidbMonitor is NodePort
    2. Add EnvVar sort to avoid the monitor rendering different results from the same TidbMonitor spec
    3. Fix the problem that the TidbMonitor LoadBalancer IP is not used (#1962, @Yisaer)
  • Make tidb-initializer support TLS (#1931, @weekface)
    1. Fix the problem that Advanced StatefulSet cannot work with webhook
    2. Change the Reaction for the Down State TiKV pod during deleting request in webhook from admit to reject (#1963, @Yisaer)
  • Fix the drainer installation error when drainerName is set (#1961, @DanielZhangQD)
  • Fix some TiKV configuration keys in toml (#1887, @aylei)
  • Support using a remote directory as data source for tidb-lightning (#1629, @aylei)
  • Add the API document and a script that generates documentation (#1945, @Yisaer)
  • Add the tikv-importer chart (#1910, @shonge)
  • Fix the Prometheus scrape config issue while TLS is enabled (#1919, @weekface)
  • Enable TLS between TiDB components (#1870, @weekface)
  • Fix the timeout error when .Values.admission.validation.pods is true during the TiKV upgrade (#1875, @Yisaer)
  • Enable TLS for MySQL clients (#1878, @weekface)
  • Fix the bug which would cause broken TiDB image property (#1860, @Yisaer)
  • TidbMonitor would use its namespace for the targetRef if it is not defined (#1834, @Yisaer)
  • Support starting tidb-server with --advertise-address parameter (#1859, @LinuxGit)
  • Backup/Restore: support configuring TiKV GC life time (#1835, @LinuxGit)
  • Support no secret for S3/Ceph when the OIDC authentication is used (#1817, @tirsen)
    1. Change the setting from the previous admission.hookEnabled.pods to the admission.validation.pods
    2. Change the setting from the previous admission.hookEnabled.statefulSets to the admission.validation.statefulSets
    3. Change the setting from the previous admission.hookEnabled.validating to the admission.validation.pingcapResources
    4. Change the setting from the previous admission.hookEnabled.defaulting to the admission.mutation.pingcapResources
    5. Change the setting from the previous admission.failurePolicy.defaulting to the admission.failurePolicy.mutation
    6. Change the setting from the previous admission.failurePolicy.* to the admission.failurePolicy.validation (#1832, @Yisaer)
  • Enable TidbCluster defaulting mutation by default which is recommended when admission webhook is used (#1816, @Yisaer)
  • Fix a bug that TiKV fails to start while creating the cluster using CR with cluster TLS enabled (#1808, @weekface)
  • Support using prefix in remote storage during backup/restore (#1790, @DanielZhangQD)

TiDB Operator v1.1.0-beta.2 Release Notes

This is a pre-release of v1.1.0, which focuses on the usability, extensibility and security of TiDB Operator. While we encourage usage in non-critical environments, it is NOT recommended to use this version in critical environments.

Changes since v1.1.0-beta.1

Action Required

  • --default-storage-class-name and --default-backup-storage-class-name are abandoned, and the storage class defaults to Kubernetes default storage class right now. If you have set default storage class different than Kubernetes default storage class, please set them explicitly in your TiDB cluster helm or YAML files. (#1581, @cofyc)

Other Notable Changes

  • Allow users to configure affinity and tolerations for Backup and Restore. (#1737, @Smana)
  • Allow AdvancedStatefulSet and Admission Webhook to work together. (#1640, @Yisaer)
  • Add a basic deployment example of managing TiDB cluster with custom resources only. (#1573, @aylei)
  • Support TidbCluster Auto-scaling feature based on CPU average utilization load. (#1731, @Yisaer)
  • Support user-defined TiDB server/client certificate (#1714, @weekface)
  • Add an option for tidb-backup chart to allow reusing existing PVC or not for restore (#1708, @mightyguava)
  • Add resources, imagePullPolicy and nodeSelector field for tidb-backup chart (#1705, @mightyguava)
  • Add more SANs (Subject Alternative Name) to TiDB server certificate (#1702, @weekface)
  • Support automatically migrating existing Kubernetes StatefulSets to Advanced StatefulSets when AdvancedStatfulSet feature is enabled (#1580, @cofyc)
  • Fix the bug in admission webhook which causes PD pod deleting error and allow the deleting pod to request for PD and TiKV when PVC is not found. (#1568, @Yisaer)
  • Limit the restart rate for PD and TiKV - only one instance would be restarted each time (#1532, @Yisaer)
  • Add default ClusterRef namespace for TidbMonitor as the same as it is deployed and fix the bug that TidbMonitor's Pod can't be created when Spec.PrometheusSpec.logLevel is missing. (#1500, @Yisaer)
  • Refine logs for TidbMonitor and TidbInitializer controller (#1493, @aylei)
  • Avoid unnecessary updates to Service and Deployment of discovery (#1499, @aylei)
  • Remove some update events that are not very useful (#1486, @weekface)

TiDB Operator v1.1.0-beta.1 Release Notes

This is a pre-release of v1.1.0, which focuses on the usability, extensibility and security of TiDB Operator. While we encourage usage in non-critical environments, it is NOT recommended to use this version in critical environments.

Changes since v1.0.0

Action Required

  • ACTION REQUIRED: Add the timezone support for all charts (#1122, @weekface).

    For the tidb-cluster chart, we already have the timezone option (UTC by default). If the user does not change it to a different value (for example: Aisa/Shanghai), all Pods will not be recreated. If the user changes it to another value (for example: Aisa/Shanghai), all the related Pods (add a TZ env) will be recreated (rolling update).

    Regarding other charts, we don't have a timezone option in their values.yaml. We add the timezone option in this PR. No matter whether the user uses the old values.yaml or the new values.yaml, all the related Pods (add a TZ env) will not be recreated (rolling update).

    The related Pods include pump, drainer, discovery, monitor, scheduled backup, tidb-initializer, and tikv-importer.

    All images' time zone maintained by tidb-operator is UTC. If you use your own images, you need to make sure that the time zone inside your images is UTC.

Other Notable Changes

  • Support backup to S3 with Backup & Restore (BR) (#1280, @DanielZhangQD)
  • Add basic defaulting and validating for TidbCluster (#1429, @aylei)
  • Support scaling in/out with deleted slots feature of advanced StatefulSets (#1361, @cofyc)
  • Support initializing the TiDB cluster with TidbInitializer Custom Resource (#1403, @DanielZhangQD)
  • Refine the configuration schema of PD/TiKV/TiDB (#1411, @aylei)
  • Set the default name of the instance label key for tidbcluster-owned resources to the cluster name (#1419, @aylei)
  • Extend the custom resource TidbCluster to support managing the Pump cluster (#1269, @aylei)
  • Fix the default TiKV-importer configuration (#1415, @aylei)
  • Expose ephemeral-storage in resource configuration (#1398, @aylei)
  • Add e2e case of operating tidb-cluster without helm (#1396, @aylei)
  • Expose terraform Aliyun ACK version and specify the default version to '1.14.8-aliyun.1' (#1284, @shonge)
  • Refine error messages for the scheduler (#1373, @weekface)
  • Bind the cluster-role system:kube-scheduler to the service account tidb-scheduler (#1355, @shonge)
  • Add a new CRD TidbInitializer (#1391, @aylei)
  • Upgrade the default backup image to pingcap/tidb-cloud-backup:20191217 and facilitate the -r option (#1360, @aylei)
  • Fix Docker ulimit configuring for the latest EKS AMI (#1349, @aylei)
  • Support sync pump status to tidb-cluster (#1292, @shonge)
  • Support automatically creating and reconciling the tidb-discovery-service for tidb-controller-manager (#1322, @aylei)
  • Make backup and restore more universal and secure (#1276, @onlymellb)
  • Manage PD and TiKV configurations in the TidbCluster resource (#1330, @aylei)
  • Support managing the configuration of tidb-server in the TidbCluster resource (#1291, @aylei)
  • Add schema for configuration of TiKV (#1306, @aylei)
  • Wait for the TiDB host:port to be opened before processing to initialize TiDB to speed up TiDB initialization (#1296, @cofyc)
  • Remove DinD related scripts (#1283, @shonge)
  • Allow retrieving credentials from metadata on AWS and GCP (#1248, @gregwebs)
  • Add the privilege to operate configmap for tidb-controller-manager (#1275, @aylei)
  • Manage TiDB service in tidb-controller-manager (#1242, @aylei)
  • Support the cluster-level setting for components (#1193, @aylei)
  • Get the time string from the current time instead of the Pod name (#1229, @weekface)
  • Operator will not resign the ddl owner anymore when upgrading tidb-servers because tidb-server will transfer ddl owner automatically on shutdown (#1239, @aylei)
  • Fix the Google terraform module use_ip_aliases error (#1206, @tennix)
  • Upgrade the default TiDB version to v3.0.5 (#1179, @shonge)
  • Upgrade the base system of Docker images to the latest stable (#1178, @AstroProfundis)
  • tkctl get TiKV now can show store state for each TiKV Pod (#916, @Yisaer)
  • Add an option to monitor across namespaces (#907, @gregwebs)
  • Add the STOREID column to show the store ID for each TiKV Pod in tkctl get TiKV (#842, @Yisaer)
  • Users can designate permitting host in chart values.tidb.permitHost (#779, @shonge)
  • Add the zone label and reserved resources arguments to kubelet (#871, @aylei)
  • Fix an issue that kubeconfig may be destroyed in the apply phrase (#861, @cofyc)
  • Support canary release for the TiKV component (#869, @onlymellb)
  • Make the latest charts compatible with the old controller manager (#856, @onlymellb)
  • Add the basic support of TLS encrypted connections in the TiDB cluster (#750, @AstroProfundis)
  • Support tidb-operator to spec nodeSelector, affinity and tolerations (#855, @shonge)
  • Support configuring resources requests and limits for all containers of the TiDB cluster (#853, @aylei)
  • Support using Kind (Kubernetes IN Docker) to set up a testing environment (#791, @xiaojingchen)
  • Support add-hoc data source to be restored with the tidb-lightning chart (#827, @tennix)
  • Add the tikvGCLifeTime option (#835, @weekface)
  • Update the default backup image to pingcap/tidb-cloud-backup:20190828 (#846, @aylei)
  • Fix the Pump/Drainer data directory to avoid potential data loss (#826, @aylei)
  • Fix the issue thattkctl ouputs nothing with the -oyaml or -ojson flag and support viewing details of a specific Pod or PV, also improve the output of the tkctl get command (#822, @onlymellb)
  • Add recommendations options to mydumper: -t 16 -F 64 --skip-tz-utc (#828, @weekface)
  • Support zonal and multi-zonal clusters in deploy/gcp (#809, @cofyc)
  • Fix ad-hoc backup when the default backup name is used (#836, @DanielZhangQD)
  • Add the support for tidb-lightning (#817, @tennix)
  • Support restoring the TiDB cluster from a specified scheduled backup directory (#804, @onlymellb)
  • Fix an exception in the log of tkctl (#797, @onlymellb)
  • Add the hostNetwork field in PD/TiKV/TiDB spec to make it possible to run TiDB components in host network (#774, @cofyc)
  • Use mdadm and RAID rather than LVM when it is available on GKE (#789, @gregwebs)
  • Users can now expand cloud storage PV dynamically by increasing the PVC storage size (#772, @tennix)
  • Support configuring node image types for PD/TiDB/TiKV node pools (#776, @cofyc)
  • Add a script to delete unused disk for GKE (#771, @gregwebs)
  • Support binlog.pump.config and binlog.drainer.config configurations for Pump and Drainer (#693, @weekface)
  • Prevent the Pump progress from exiting with 0 if the Pump becomes offline (#769, @weekface)
  • Introduce a new helm chart, tidb-drainer, to facilitate multiple Drainers management (#744, @aylei)
  • Add the backup-manager tool to support backing up, restoring, and cleaning backup data (#694, @onlymellb)
  • Add affinity to Pump/Drainer configration (#741, @weekface)
  • Fix the TiKV scaling failure in some cases after TiKV failover (#726, @onlymellb)
  • Fix error handling for UpdateService (#718, @DanielZhangQD)
  • Reduce e2e run time from 60 m to 20 m (#713, @weekface)
  • Add the AdvancedStatefulset feature to use advanced StatefulSet instead of Kubernetes builtin StatefulSet (#1108, @cofyc)
  • Enable auto generate certificates for the TiDB cluster (#782, @AstroProfundis)
  • Support backup to gcs (#1127, @onlymellb)
  • Support configuring net.ipv4.tcp_keepalive_time and net.core.somaxconn for TiDB and configuring net.core.somaxconn for TiKV (#1107, @DanielZhangQD)
  • Add basic e2e tests for aggregated apiserver (#1109, @aylei)
  • Add the enablePVReclaim option to reclaim PV when tidb-operator scales in TiKV or PD (#1037, @onlymellb)
  • Unify all S3 compliant storage to support backup and restore (#1088, @onlymellb)
  • Set podSecuriyContext to nil by default (#1079, @aylei)
  • Add tidb-apiserver in the tidb-operator chart (#1083, @aylei)
  • Add new component TiDB aggregated apiserver (#1048, @aylei)
  • Fix the issue that the tkctl version does not work when the release name is un-wanted (#1065, @aylei)
  • Support pause for backup schedule (#1047, @onlymellb)
  • Fix the issue that TiDB Loadbalancer is empty in terraform output (#1045, @DanielZhangQD)
  • Fix that the create_tidb_cluster_release variable in AWS terraform script does not work (#1062, @aylei)
  • Enable ConfigMapRollout by default in the stability test (#1036, @aylei)
  • Migrate to use app/v1 and do not support Kubernetes before 1.9 anymore (#1012, @Yisaer)
  • Suspend the ReplaceUnhealthy process for AWS TiKV auto-scaling-group (#1014, @aylei)
  • Change the tidb-monitor-reloader image to pingcap/tidb-monitor-reloader:v1.0.1 (#898, @qiffang)
  • Add some sysctl kernel parameter settings for tuning (#1016, @tennix)
  • Support maximum retention time backups for backup schedule (#979, @onlymellb)
  • Upgrade the default TiDB version to v3.0.4 (#837, @shonge)
  • Fix values file customization for tidb-operator on Aliyun (#971, @DanielZhangQD)
  • Add the maxFailoverCount limit to TiKV (#965, @weekface)
  • Support setting custom tidb-operator values in terraform script for AWS (#946, @aylei)
  • Convert the TiKV capacity into MiB when it is not a multiple of GiB (#942, @cofyc)
  • Fix Drainer misconfiguration (#939, @weekface)
  • Support correctly deploying tidb-operator and tidb-cluster with customized values.yaml (#959, @DanielZhangQD)
  • Support specifying SecurityContext for PD, TiKV and TiDB Pods and enable tcp keepalive for AWS (#915, @aylei)