-
Notifications
You must be signed in to change notification settings - Fork 5.8k
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
[Data] Add retry for _sample_fragment
during ParquetDatasource._estimate_files_encoding_ratio()
#42759
Conversation
Signed-off-by: Scott Lee <[email protected]>
Signed-off-by: Scott Lee <[email protected]>
Signed-off-by: Scott Lee <[email protected]>
sample_fragment.options( | ||
scheduling_strategy=scheduling, | ||
# Retry in case of transient errors during sampling. | ||
retry_exceptions=[OSError], |
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.
Let's extract this as a variable similar to RETRY_EXCEPTIONS_FOR_META_FETCH_TASK
, so we can tune later.
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.
discussed with @scottjlee , we'll make this configurable in DataContext. Since other retry errors are not configurable either. we'll use a follow-up PR to update all.
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 feel it's okay to not put into DataContext. I don't think they are user-facing, and mostly it only should be tuned by ourselves, when debugging user job. Let's only add config to DataContext if they are expected to be set by users.
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.
But agreed to merge this PR ASAP to unblock users.
…timate_files_encoding_ratio()` (ray-project#42759)
Why are these changes needed?
Sometimes, there can be transient errors during the fragment sampling phase of
read_parquet()
, when Ray Data estimates the file encoding ratio prior to execution. We include this fragment sampling phase into a retry loop so that we automatically retry upon failure.Related issue number
Checks
git commit -s
) in this PR.scripts/format.sh
to lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/
under thecorresponding
.rst
file.