v0.3.5 Release

v0.3.5 (2021-07-08)

Full Changelog

Closed issues:

• Type of replicaType is not consistent in common project #134
• project cannot complie with kubeflow/common v0.3.3 and client-go v0.19 #132
• Bump dependency to 1.19 and upgrade client to accept Context #125
• Make training container port support multi ports #122

Merged pull requests:

• fix(expectation): more appropriate use of expectation #139 (eggiter)
• optimize the ttl of job #137 (HeGaoYuan)
• Fix dependency configuration #136 (MartinForReal)
• change type of rtype to commonv1.ReplicaType #135 (MartinForReal)
• Minor edits and fix typos in README.md #133 (terrytangyuan)

v0.3.4 release

v0.3.4 (2021-05-18)

Full Changelog

Closed issues:

• [feature] Make training container port support multi ports #122
• [chore] project cannot complie with kubeflow/common v0.3.3 and client-go v0.19 #132
• [chore] Bump dependency to 1.19 and upgrade client to accept Context #125

Merged pull requests:

• Bump k8s dependency to 1.19 and add context support (#126, @Jeffwan)
• Make ports more flexiable for customizations (#131, @xieydd)
• chore: Retire inactive owners (#128, @terrytangyuan)

v0.3.3 release

v0.3.3 (2021-03-31)

Full Changelog

Closed issues:

• [feature] Support gang scheduling configuration in common #113

Merged pull requests:

• Clean up go modules #127 (Jeffwan)
• update Volcano CRD API to v1.16 #123 (Thor-wl)
• feat(common): support gang scheduling config #116 (shinytang6)
• Add tfjob annotations to podgroup. #115 (jiangkaihua)
• feat: Add docs for API #114 (gaocegege)
• feat: Remove k8s.io/kubernetes from go mod #111 (gaocegege)

v0.3.2 release

v0.3.2 (2021-01-26)

Full Changelog

Closed issues:

• CleanPodPolicyRunning should also clean Pending pods #104
• [Release 1.1] training operator release #97
• PodGroup sync loop is missing #94
• Consider how to refactor pkg/controller.v1beta2/tensorflow #7

Merged pull requests:

• reorder job cleanup and completion time setting #108 (shaowei-su)
• CleanPodPolicyRunning cleans Pending phase pod #105 (mkkb473)
• Feat(event): record the tfjob event from the abnormal pod status #101 (whalecold)
• Add v0.3.1 release notes #95 (terrytangyuan)

v0.3.1 release

• Skip some logics for terminated job and add PodGroup reconcile loop (#93, @Jeffwan)
• Provides more flexibility for customizations (#91, @Jeffwan)
• Change expectation package log level to debug (#90, @Jeffwan)
• Bump Kubernetes dependency to 1.16.9 (#87, @Jeffwan)
• Add release notes document (#88, @terrytangyuan)

v0.3.0 release

• Simplify interfaces and remove useless interfaces (#85, @Jeffwan)
• Add scheduling related Prometheus metrics (#84, @terrytangyuan)
• Reformat code with go fmt and check this on Travis CI (#83, @terrytangyuan)
• Add Prometheus metrics for pod created/deleted/failed (#82, @terrytangyuan)
• Fix crash when ReplicaSpec.Replicas is nil (#81, @hustcat)
• Add @Jeffwan to OWNERS (#80, @Jeffwan)
• Edits on dev guide and rename it to CONTRIBUTING.md (#79, @terrytangyuan)
• Add proposal for Prometheus metrics coverage (#77, @terrytangyuan)
• Add DEVELOPMENT.md doc (#78, @Jeffwan)
• Create expectation package (#74, @Jeffwan)
• Add PodControlInterface and PodControllerRefManager (#73, @Jeffwan)
• Add Prometheus metrics for service creation (#75, @terrytangyuan)
• Update comments on the use of -ignore flag for goveralls (#76, @terrytangyuan)
• Move Pods/Services control interface to separate folder (#72, @Jeffwan)
• Import volcano api (#62, @hzxuzhonghu)

v0.2.0 release

• Add links to issues on ROADMAP.md (#69, @terrytangyuan)
• Add distributed training operator roadmap (#61, @Jeffwan)
• Remove vendor dependencies (#63, @Jeffwan)
• Make container port optional (#57, @Jeffwan)
• Support the case user down scale replicas (#58, @Jeffwan)
• Enhance maintainability of operator common module (#55, @Jeffwan)
• Deprecate glog and move to commonutil logger (#53, @Jeffwan)

v0.1.1 release

• Remove unnecessary comment on gometalinter (#50, @terrytangyuan)
• Update openapi tag for SDK generating (#49, @jinchihe)

v0.1.0 release

• [hot-fix-43] change the job status api (#44, @merlintang)
• Add readme (#42, @jian-he)
• GetDefaultContainerPortNumber to GetDefaultContainerPortName (#41, @merlintang)
• re-org api package (#39, @jian-he)
• Remove the podControl and serviceControl interfaces (#36, @jian-he)
• refine arguments of ControllerInterface.UpdateJobStatus (#35, @sperlingxx)
• Generate vendor directories (#28, @terrytangyuan)
• Add skeleton code for reconcile service (#25, @merlintang)
• Correct function names in the comment (#32, @terrytangyuan)
• Make UpdateJobConditions public to be used by custom operators (#30, @jian-he)
• Unify the label keys (#29, @jian-he)
• Add utility methods for job reconciliation (#24, @terrytangyuan)
• test: Add util/train/train_util_test.go (#26, @gaocegege)
• chore: Fix package name (#27, @gaocegege)
• Add utility methods for pod creation (#17, @terrytangyuan)
• Remove mentions of tensorflow in test job (#21, @terrytangyuan)
• Fix incorrect name for restart policy exit code (#20, @terrytangyuan)
• Update go.sum changed by golangci-lint (#18, @terrytangyuan)
• Added .gitignore file (#16, @terrytangyuan)
• Add more utility functions for job and status (#14, @jian-he)
• Added missing generated lines in go.sum (#15, @terrytangyuan)
• Added schedulingPolicy api. (#13, @k82cn)
• Define interfaces for common job operators (#12, @jian-he)
• Add Travis badge and Go report card (#9, @terrytangyuan)
• Enable Travis CI on kubeflow/common (#6, @richardsliu)
• Common job controller library (#5, @richardsliu)
• Add terrytangyuan to OWNERS (#4, @richardsliu)
• Add common types to kubeflow/common (#2, @richardsliu)
• Create LICENSE (#1, @richardsliu)