tidb-operator v1.0.0-beta.3

This release focuses on stability and ease of use. Please see CHANGELOG.md for details.

Docker images:

docker pull pingcap/tidb-operator:v1.0.0-beta.3
docker pull uhub.ucloud.cn/pingcap/tidb-operator:v1.0.0-beta.3

For user guide, please check out user guide.

tidb-operator v1.0.0-beta.2

This release still focuses on stability. Please see CHANGELOG.md for details.

Docker images:

docker pull pingcap/tidb-operator:v1.0.0-beta.2
docker pull uhub.ucloud.cn/pingcap/tidb-operator:v1.0.0-beta.2

For user guide, please check out user guide.

tidb-operator v1.0.0-beta.1-p2

This release still focuses on stability. See the detailed release notes below:

• new algorithm for scheduler HA predicate (#260)
• add tidb discovery service (#262)
• serial scheduling (#266)
• change tolerations type to an array (#271)
• start directly when where is join file (#275)
• add code coverage icon (#272)
• in values.yml, omit just the empty leaves (#273)
• charts: backup to ceph object storage (#280)
• add ClusterIDLabelKey label to TidbCluster (#279)

tidb-operator v1.0.0-beta.1-p1

This release still focuses on stability. See the detailed release notes below:

Bug fixes:

• Fix scheduler policy issue, works on kubernetes v1.10, v1.11 and v1.12 now #256

Docs:

• Proposal: add multiple statefulsets support to TiDB Operator #240
• Update roadmap #258

tidb-operator v1.0.0-beta.1

This release still focuses on stability. We've run lots of chaos tests since the last release, so there are lots of bugfix in this release. Also, we've made some minor improvements to make tidb-operator more user-friendly. See the detailed release notes below:

Bug fixes:

• Fix pd_control bug: avoid relying on PD error response text #197
• Add orphan pod cleaner #201
• Fix scheduler configuration for Kubernetes 1.12 #200
• Fix Grafana configuration #206
• Fix pd failover bug: scale out directly when failover occurs #217
• Refactor PD failover #211
• Refactor tidb_cluster_control logic #215
• Fix upgrade logic: avoid updating pd/tikv/tidb simutaneously #234
• Fix PD control logic: get member/store before delete member/store and fix member id parse error #245
• Fix documents errors #213
• Fix backup and restore script bug #251 #254 #255
• Fix GKE multiple availablity zones deployment PD disk scheduling bug #248

Minor improvements:

• Add Kubernetes 1.12 local DinD scripts #195
• Bump default TiDB to v2.1.0 #212
• Release tidb-operator/tidb-cluster charts #216
• Add connection timeout for TiDB password setter job #219
• Separate ad-hoc backup and restore to another chart #227
• Add compiler version info to tidb-operator binary #237
• Allow specifying TiDB service LoadBalancer IP #246
• Expose TiKV cpu/memory related configuration to vlaues.yaml #252

tidb-operator v1.0.0-beta.0

This release focuses on stability, the unit tests coverage is greatly improved (80%), some code are refactored during adding unit test case, and basic chaos test cases are added, many bugs are fixed in this version. See the detailed release notes below:

• Introduce basic chaos testing
• Improve unit test coverage #179 #181 #182 #184 #190 #192 #194
• Add default value for log-level of PD/TiKV/TiDB #185
• Fix PD connection timeout issue for DinD environment #186
• Fix monitor configuration #193
• Fix document Helm client version requirement #175
• Keep scheduler name consistent in chart #188
• Remove unnecessary warning message when volumeName is empty #177
• Migrate to Go 1.11 module #178
• Add user guide #187

tidb-operator v1.0.0-alpha

filter out master by role without relying on name (#172)

tidb-operator v0.4.0

Notable changes:

• Extend Kubernetes built-in scheduler for TiDB data awareness pod scheduling #145
• Restore backup data from GCS bucket #160
• Set password for TiDB when a TiDB cluster is first deployed #171

Minor changes and bug fix:

• Update roadmap for the following two months #166
• Add more unit tests #169
• E2E test with multiple clusters #162
• E2E test for meta info synchronization #164
• Add TiDB failover limit #163
• Synchronize PV reclaim policy early to persist data #169
• Use helm release name as instance label #168 (break change)
• Fix local PV setup script #172

tidb-operator v0.3.1

Minor changes:

• paramertize the serviceAccount #116 #111
• bump tidb to v2.0.7 & allow user specified config files #121
• remove binding mode for GKE pd-ssd storageclass #130
• modified placement of tidb_version #125
• update google-kubernetes-tutorial.md #105
• remove redundant creation statement of namespace tidb-operator-e2e #132
• update the label name of app in local dind documentation #136
• remove noisy events #131
• marketplace #123 #135
• change monitor/backup/binlog pvc labels #143
• TiDB readiness probes #147
• Add doc on how to provision kubernetes on aws #71
• Add imagePullPolicy support #152
• separation startup scripts and application config from yaml files #149
• update marketplace for our open source offering #151
• add validation to crd #153
• marketplace: use the Release.Name #157

Bug fix:

• fix parallel upgrade bug #118
• fix wrong parameter AGRS to ARGS #114
• can't recover after a upgrade failed #120
• scale in when store id match #124
• pd can't scale out if not all members are ready #142
• podLister and pvcLister usages are wrong #158

tidb-operator v0.3.0

Notable changes:

• Add fullbackup support
• Add TiDB Binlog support
• Add graceful upgrade feature
• Allow monitor data to be persistent