Closed alculquicondor closed 1 year ago
/kind feature
/assign
@mimowo We have completed upgrading the kubernetes dependencies in #502. It may help you implement suspend semantics.
@tenzen-y @alculquicondor you may want to look at the WIP implementation (tested manually) here: https://github.com/kubeflow/mpi-operator/pull/511. Any early feedback is welcome.
Probably, we can close this issue.
/close
@alculquicondor: Closing this issue.
The semantics should be similar to that of k8s Job.
And this will pave the work for the training-operator (https://github.com/kubeflow/training-operator/issues/1519)