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.

Sign up for GitHub

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

Jump to bottom

Fix sky spot launch TPU #1249

Merged
merged 2 commits into from
Oct 18, 2022
Merged

Fix sky spot launch TPU #1249

merged 2 commits into from
Oct 18, 2022

Conversation

infwinston
Copy link
Member

Now sky spot launch a TPU VM will fail due to gcloud not fully setup on the spot controller.
This PR fixes it by uploading ~/.config/gcloud/active_config.

@concretevitamin
Copy link
Member

Great catch @infwinston. Can we add a smoke test?

@infwinston
Copy link
Member Author

Just added!

concretevitamin
concretevitamin approved these changes Oct 14, 2022
View reviewed changes
Copy link
Member

@concretevitamin concretevitamin left a comment

Choose a reason for hiding this comment

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

LGTM - with this, does the user example run successfully as a managed spot job?

tests/test_smoke.py
'sleep 600', # TPU takes a while to launch
f's=$(sky spot status); printf "$s"; echo; echo; printf "$s" | grep {name} | head -n1 | grep "RUNNING\|SUCCEEDED"',
],
f'sky spot cancel -y -n {name}',
Copy link
Member

Choose a reason for hiding this comment

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

We don't need a tear down command for managed spot jobs.

Copy link
Member Author

Choose a reason for hiding this comment

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

Oh this smoke test will pass if the job changed from STARTING to RUNNING. In this case, we can actually kill the job to save some costs. But we can remove it if we want as the job is short.

Copy link
Member

Choose a reason for hiding this comment

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

Oh I thought it was short - let's still keep it then.

@infwinston
Copy link
Member Author

LGTM - with this, does the user example run successfully as a managed spot job?

Yes, it does run successfully now.

@Michaelvll
Copy link
Collaborator

This may change a bit of the authentication of the GCP account. Please run smoke tests before merging the PR. ; )

@infwinston
Copy link
Member Author

infwinston commented Oct 18, 2022 •
edited
Loading

On RISE server, smoke test passed except for test_spot_storage. Offline discussion with @Michaelvll suggests this is not related to the PR and I file one issue at #1266. I'm merging this change now.

@infwinston infwinston merged commit 17a591e into master Oct 18, 2022
@infwinston infwinston deleted the fix-spot-tpu branch October 18, 2022 03:34
ewzeng pushed a commit to ewzeng/skypilot that referenced this pull request Oct 24, 2022
@infwinston @ewzeng
Fix sky spot launch TPU (skypilot-org#1249)
14047c8 
* Add active config

* Add smoke test
ewzeng pushed a commit to ewzeng/skypilot that referenced this pull request Oct 24, 2022
@infwinston @ewzeng
Fix sky spot launch TPU (skypilot-org#1249)
f515bd2 
* Add active config

* Add smoke test
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@concretevitamin concretevitamin concretevitamin approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@infwinston @concretevitamin @Michaelvll