Skip to content

[query] Fix text export part file overwrite due to speculation#11910

Merged
danking merged 1 commit intohail-is:mainfrom
chrisvittal:lowering/fix-speculation-overwrite
Jun 14, 2022
Merged

[query] Fix text export part file overwrite due to speculation#11910
danking merged 1 commit intohail-is:mainfrom
chrisvittal:lowering/fix-speculation-overwrite

Conversation

@chrisvittal
Copy link
Collaborator

We do not make guarantees about the exact contents of a folder after
a parallel export. We do guarantee that all data is in the folder
however, so we must make it speculation safe. We do so by adding a UUID
to each part file we write for in lowered text export.

@tpoterba
Copy link
Contributor

looks good!

We do not make guarantees about the exact contents of a folder after
a parallel export. We do guarantee that all data is in the folder
however, so we must make it speculation safe. We do so by adding a UUID
to each part file we write for in lowered text export.
@chrisvittal chrisvittal force-pushed the lowering/fix-speculation-overwrite branch from a33de55 to 70dbb3c Compare June 14, 2022 16:47
@danking danking merged commit 4ea7537 into hail-is:main Jun 14, 2022
chrisvittal added a commit to chrisvittal/hail that referenced this pull request Jun 16, 2022
This fixes an issue introduced by hail-is#11910, where extra part files may be
created by speculative execution. This is an issue for parallel export
where we did not delete any files.

Now we list the output directory, and compare it to the proper list of
files. The paths that are not in the list of 'good' paths are deleted.
chrisvittal added a commit to chrisvittal/hail that referenced this pull request Jun 17, 2022
This fixes an issue introduced by hail-is#11910, where extra part files may be
created by speculative execution. This is an issue for parallel export
where we did not delete any files.

Now we list the output directory, and compare it to the proper list of
files. The paths that are not in the list of 'good' paths are deleted.
danking pushed a commit that referenced this pull request Jun 17, 2022
This fixes an issue introduced by #11910, where extra part files may be
created by speculative execution. This is an issue for parallel export
where we did not delete any files.

Now we list the output directory, and compare it to the proper list of
files. The paths that are not in the list of 'good' paths are deleted.
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.

3 participants