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Outdated suggestions cannot be applied.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.
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 serve ft nightly #19125
Add serve ft nightly #19125
Changes from 3 commits
f1c4b97
4e28f41
78a9426
e9dbcfd
70ec445
4171c0c
6fd37e6
626b859
File filter
Filter by extension
Conversations
Jump to
There are no files selected for viewing
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.
i think previously we're not really getting these tokens in make_kv_store, so adding it at client level with actual s3 integration.
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.
boto should fall back to environ if we pass in None, so this is not necessary.
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.
bump on this
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.
Hmm, I think we should still fully test that path where the entire cluster goes down. If we just test the controller being killed there may be some hidden assumptions about state in the GCS or something.
Can you just use
cluster_utils
and blow away the whole cluster and create a new one?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.
I would like to completely start new ray cluster if possible as well, but think cluster_utils is for local tests, is there an equivalent on product to blow away current cluster and start new one ?
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.
Hmm I'm not sure. cluster_utils would be a good first step, it at least clears all process-level state
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.
Ask internally on slack if there's a best way to do this? Worst case you just use the SDK to kill the cluster and start a new one.
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.
cluster_utils will run much faster though and it'll be nice to be able to run it locally for testing. if possible maybe make one test with both "backends" and start with just cluster_utils?
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.
i asked a few product folks, our SDK has start / terminate cluster APIs but the issues is current releaser/e2e.py is assuming it runs a single script as if on laptop, in single ray cluster. Doing this in our test script might not work well and most feasible way is probably extend existing test script's yaml field to extend a multi-cluster setup.
For now im trying to work with multiple local clusters on product, which should terminate quick but also give us more coverage compare to current revision.