-
Notifications
You must be signed in to change notification settings - Fork 32
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add Success Policy to JobSet API #124
Conversation
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: danielvegamyhre The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
9531f84
to
49f5346
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
/lgtm
But probably should leave final approval for @ahg-g.
/hold just in case it merges by mistake |
Sorry about that. Forgot that /lgtm will cause a merge since you and @danielvegamyhre are approvers. |
/label tide/merge-method-squash |
Did some testing! For others interested in doing this, you need to install the development version to your cluster and the module from the main branch reference:
And then testing - I tested a job name that didn't exist, this seems like the correct result! 2023-05-19T23:57:34Z DEBUG controller-runtime.webhook.webhooks wrote response {"webhook": "/validate-jobset-x-k8s-io-v1alpha1-jobset", "code": 403, "reason": "invalid replicatedJob name 'broker' does not appear in .spec.ReplicatedJobs", "UID": "0becc279-1e3a-48db-8328-b70dc007406a", "allowed": false} And then the actual use case - a broker job in the jobset successfully finishing:
You can see how we are using it here. We actually have a design for the operator now that the workers do exit when the broker leader exits, but this is nice because it's mapping that same logic into Kubernetes too. I think once I've tested everything, the decision point will likely come down to being able to test this in a more production sense. The pros of this design are separation of the worker / broker logic into separate scripts. The cons are the longer names of the pods and (at least in this setup) what appears to be a bit of slowness sometimes. But I'm going to chock that up to my setup, and will assume when it's added proper it will be on par with the current objects in batch! The last thing to test is #136, and one thing I'm realizing is we would want to make sure that the label selector is transparent to be able to potentially apply it to other objects in the cluster (e.g., for the Flux Operator, a one off container service). |
Fixes #82
Fixes #120