Support cluster domain for MPI HostFile#704
Merged
google-oss-prow[bot] merged 1 commit intokubeflow:masterfrom Aug 14, 2025
Merged
Support cluster domain for MPI HostFile#704google-oss-prow[bot] merged 1 commit intokubeflow:masterfrom
google-oss-prow[bot] merged 1 commit intokubeflow:masterfrom
Conversation
Signed-off-by: Yuki Iwai <yuki.iwai.tz@gmail.com>
Member
Author
|
/assign @terrytangyuan |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: terrytangyuan The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
| fs.IntVar(&s.ControllerRateLimit, "controller-queue-rate-limit", 10, "Rate limit of the controller events queue .") | ||
| fs.IntVar(&s.ControllerBurst, "controller-queue-burst", 100, "Maximum burst of the controller events queue.") | ||
|
|
||
| fs.StringVar(&s.ClusterDomain, "cluster-domain", "", ` |
Member
There was a problem hiding this comment.
@tenzen-y Do we need this for MPI plugin in KF Trainer ?
Member
Author
There was a problem hiding this comment.
I don't think so since Trainer does not support <PodName>.<JobName>.<namespace>.svc format. The Trainer supports only <PodName>.<JobName>. In case of Trainer, the cluster domain is automatically considered, immediately.
This was referenced Oct 29, 2025
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
I added
--cluster-domainoption to add the cluster domain when MPIOperator constructs MPI HostFile.When it is specified the HostFile domains are built with
<pod-name>.<mpi-job-name>.<namespace>.svc.<cluster-domain>format. Otherwise,<pod-name>.<mpi-job-name>.<namespace>.svcis used the same as previously.Background
In #453, we added
<namespace>.svcsuffix to address a special network environment. But, it is not an ideal "headless service" endpoint name format for the generic network environment, wheresvc.<cluster-domain>is configured by the kubelet. As a result, there are many DNS name resolutions trying until hitting the domain w/out a cluster domain. That causes the Launcher Job to be recreated many times due to the failure of DNS name resolution.This is quite impacted in the runLauncherAsWorker environment since Launcher is launched very fast before the svc. is set up.