Skip to content
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

Set up JobSet failure policy on workloads #308

Open
wants to merge 1 commit into
base: main
Choose a base branch
from

Conversation

IrvingMg
Copy link
Collaborator

@IrvingMg IrvingMg commented Jan 3, 2025

Fixes / Features

  • Handling exit codes on Workloads with JobSet failure policy.

Testing / Documentation

Testing details.

  • [ y/n ] Tests pass
  • [ y/n ] Appropriate changes to documentation are included in the PR

@IrvingMg IrvingMg force-pushed the feature/set-up-jobset-failure-policy branch from e5f75ed to 63738ec Compare January 3, 2025 14:00
@@ -445,6 +453,27 @@ def workload_create(args) -> None:
if return_code != 0:
xpk_exit(return_code)

failure_policy_rules = ''
pod_failure_policy = ''
if args.restart_on_user_code_failure:
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is great to use failure policy!! Looking forward to this change.

I named this argument poorly -- its intent is to allow for the workload to restart additionally on user code failures when set. The range of exit codes this should specify is anything that a user would specify like 1,2,...

Some system level exit codes that should always restart the job are 42, 143. Probably should ask GKE folks what other system level exit codes exist. These should always restart the job and should be set / or assumed in the pod failure policy to restart by default.


For background some customers depend on xpk to restart the job in system failure cases by default and avoid user errors (like code typos) from restarting their workload. So this is the default case.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

To be honest, I don't think anyone actually wants a user error to restart the job so the ideal choice in my opinion is to remove the "args.restart_on_user_code_failure" argument and just "restart the job in system failure cases by default and avoid user errors (like code typos) from restarting their workload". Since now with the failure policy support we can do this.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for the information. Then, to sum up:

  1. Default behavior should be: restart on exit codes 42 and 143.
  2. Remove restart-on-user-code-failure flag but allow users to pass additional exit codes for restarting workload.

Is that correct? And how is the user going to pass these additional exit codes, should we have another flag for that?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@Obliviour can you confirm those are the changes needed?

@IrvingMg IrvingMg force-pushed the feature/set-up-jobset-failure-policy branch 3 times, most recently from d8052eb to 7299390 Compare January 7, 2025 09:45
@IrvingMg IrvingMg marked this pull request as ready for review January 7, 2025 10:16
@IrvingMg IrvingMg force-pushed the feature/set-up-jobset-failure-policy branch from 7299390 to f6da574 Compare January 9, 2025 13:49
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants