kubeflow / sdk

67%

LAST BUILD ON BRANCH main

branch: SELECT

CHANGE BRANCH

x

Sync Branches

• No branch selected
• 0.1.0
• 0.1.0rc1
• 0.2.0
• 0.2.1
• add-coverage-badge
• add-dependabot
• add-labels-annotations-support#87
• add-ok-to-test-workflow
• add-pre-commit-ci
• add-proposal-title
• add-sdk-docs-workflow
• add-spark-to-roadmap
• agent.md
• allow-trainjob-naming
• auto-release
• backend-public
• cache
• check
• cherry-pick-103-to-release-0.1
• cherry-pick-154-to-release-0.2
• cherry-pick-157-to-release-0.2
• cherry-pick-162-to-release-0.2
• cherry-pick-166-to-release-0.2
• cherry-pick-168-to-release-0.2
• cherry-pick-177-to-release-0.2
• chore/fix-welcome-workflow
• ci-permissions
• coveralls-parallel
• custom-trainer-container
• dependabot/github_actions/actions/checkout-6
• dependabot/github_actions/actions/download-artifact-6
• dependabot/github_actions/actions/download-artifact-7
• dependabot/github_actions/actions/github-script-8
• dependabot/github_actions/actions/setup-python-6
• dependabot/github_actions/actions/stale-10
• dependabot/github_actions/actions/upload-artifact-6
• dependabot/github_actions/amannn/action-semantic-pull-request-6.1.1
• dependabot/github_actions/softprops/action-gh-release-2
• dev-testing
• dict-unpack
• docker-backend
• docs/update-agents-trainer-section
• drop-py39-support
• export-custom-trainer-container
• feat/171-add-initializer-support-container-backend
• feat/add-mig-device-support
• feat/container-gpu-passthrough
• feat/initializers/s3
• feat/kep-spark-client
• feat/logging-system-v2
• feat/lora-support
• feat/wait-for-pod-logs
• feature/customtrainer-python-file
• feature/support-multiple-pip-index-urls-clean
• fix-copy-get-packages
• fix-first-time-ci
• fix-image-runtime
• fix-pip-user-issue
• fix-s3-initializer
• fix/filter-duplicate-pods-in-get-job
• fix/import-trainer
• fix/namespace-trainingruntime-list
• from-exception
• get-best-results
• get-job-logs-opt
• get-job-logs-refactor
• get-runtime-packages
• hatchling-package
• hpo-new-apis
• hpo-proposal
• hpo-support
• image-issue
• install-as-user
• kubeflow-hub-api
• kubernetes-backend
• latest-news
• local-exec-proposal
• local-exec-readme
• localprocess-backend-test
• main
• move-dev-to-deps
• move-pyproject
• nominate-kramaranya
• optimizer-container
• osamaahmed17/195
• osamaahmed17/203
• podman-backend
• pr-01
• pr-02
• pr-03
• pr-05
• pr-06
• pr-197
• pr-titles-check
• prerelease-update
• remove-accelerator-type
• remove-pip-user-from-local
• roadmap
• ruff-config
• runtime-framework-labels
• skip-release-on-forks
• training-backends
• unit-test
• update-approvers
• update-contrib
• update-e2e
• update-readme
• update-readme-hpo
• update-roadmap
• update-sdk-diagram
• urllib3-update
• wait-for-job-api
• welcome-ci
• wf-ok-to-test

Build # 20780172530

Build Type

push

github

Committed by web-flow

Commit Message

feat(trainer): add get_job_events API to retrieve TrainJob events (#220)

* feat(trainer): add get_job_events() API to retrieve TrainJob events

- Add Event dataclass to represent Kubernetes events with fields: involved_object_kind, involved_object_name, message, reason, event_time, first_occurrence, last_occurrence, count, and type
- Implement get_job_events() method in TrainerClient to provide users with visibility into TrainJob state changes and pod events
- Add abstract method get_job_events() to RuntimeBackend base class
- Implement get_job_events() for KubernetesBackend with proper event filtering and sorting
- Implement get_job_events() for LocalProcessBackend and ContainerBackend (return empty list as events are Kubernetes-specific)
- Add comprehensive unit tests for the new functionality

This addresses issue #196 by providing an API for users to check TrainJob events, giving additional clarity about the state when logs alone are insufficient.

Related-To: #196
Signed-off-by: Aditya Shirsatrao <adityashirsatrao007@gmail.com>

* Update kubeflow/trainer/backends/kubernetes/backend.py

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Signed-off-by: Aditya <adityashirsatrao007@gmail.com>

* Update kubeflow/trainer/backends/kubernetes/backend.py

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Signed-off-by: Aditya <adityashirsatrao007@gmail.com>

* fix: Error handling consistent

Signed-off-by: sksingh2005 <shashanksgh3@gmail.com>

* fix: Error handling consistent

Signed-off-by: sksingh2005 <shashanksgh3@gmail.com>

* Pre-commit test failure fix

Signed-off-by: sksingh2005 <shashanksgh3@gmail.com>

* Args for docstring in Event class

Signed-off-by: sksingh2005 <shashanksgh3@gmail.com>

* test mock events API and align timeout handling

Signed-off-by: sksingh2005 <shashanksgh3@gmail.com>

* only handle timeout error from the list_namespaced_event() API

Signed-off-by: sksingh2005 <shashanksgh3@gmail.com>
... (continued)

Run Details

48 of 81 new or added lines in 7 files covered. (59.26%)

1 existing line in 1 file now uncovered.

2599 of 3902 relevant lines covered (66.61%)

1.33 hits per line