kubeflow / trainer / 21118692202

51%

push

github

fix: fix resourcePerNode override not applied with Volcano scheduler (#2982)

* volcano integeration for resourcePerNode override in trainjob

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

* Fixed the issue where PodGroup resources weren't scaling with spec.trainer.resourcesPerNode when specified on a TrainJob

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

* limits removed

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

* maintains consistency across all MLPolicy plugins

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

* Fix pre-commit formatting

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

* Unit test updates for mpi and torch pluggins

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

* framework unit test prblm solved

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

* fix:All MLPolicy Plugins update SinglePodRequests from TrainJob.spec.trainer.resourcesPerNode

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

* Refactor ResourcesPerNode handling to use PodRequest helper for correct scaling with init/sidecar containers

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

* refactor: simplify ResourcesPerNode by modifying jobSetTemplateSpec directly

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

* test: modify TrainJob resources in coscheduling test to verify node container override

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

* Refactor: remove duplicate resource logic from plugins

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

---------

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

7 of 7 new or added lines in 1 file covered. (100.0%)

1243 of 2419 relevant lines covered (51.38%)

0.61 hits per line