Skip to content

Conversation

@allieychen
Copy link
Contributor

@allieychen allieychen commented Feb 14, 2019

Fix one bug for annotation. The operations could get killed because of the watchdog file is stale even though no cancellation is made. The watchdog file was updated only when waiting for one operation to be done. This fix handles the following two failing cases:

  • If there are a lot of PAPI calls, some operations might already start running VEP (after loading cache) while we are still submitting other operations, the stale watchdog file will get this operation to be killed.
  • If there are a lot of consecutive done operations, the watchdog file won't be updated in time.

Tested: unit tests & integration tests & manually ran one annotation test.

@allieychen allieychen requested a review from nmousavi February 14, 2019 15:11
@coveralls
Copy link

coveralls commented Feb 14, 2019

Pull Request Test Coverage Report for Build 1562

  • 7 of 17 (41.18%) changed or added relevant lines in 2 files are covered.
  • 1 unchanged line in 1 file lost coverage.
  • Overall coverage decreased (-0.02%) to 88.958%

Changes Missing Coverage Covered Lines Changed/Added Lines %
gcp_variant_transforms/libs/annotation/vep/vep_runner.py 2 3 66.67%
gcp_variant_transforms/transforms/annotate_files.py 5 14 35.71%
Files with Coverage Reduction New Missed Lines %
gcp_variant_transforms/transforms/annotate_files.py 1 45.16%
Totals Coverage Status
Change from base Build 1548: -0.02%
Covered Lines: 7041
Relevant Lines: 7915

💛 - Coveralls

@kemp-google
Copy link

@allieychen Can you put the commentary about why this change is being made in the commit itself? It's much better to have information like this in the commit object rather than the PR since the PR is not a permanent part of the record.

Copy link
Contributor

@nmousavi nmousavi left a comment

Choose a reason for hiding this comment

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

Can we separate file update from runners logic, perhaps by running an updater in different thread?
I don't like the fact that update is done in multiple places, it's prone to bug and ugly. Any future change that adds a costly processing step here would require another manual update (similar to what you added in this PR), and it can be easily missed out.

@allieychen
Copy link
Contributor Author

Thanks for the review! I have moved updating the watchdog file in another thread. PTAL :)

@allieychen allieychen force-pushed the vep-bug branch 2 times, most recently from 2309da2 to 779be0b Compare February 20, 2019 18:34
Copy link
Contributor

@nmousavi nmousavi left a comment

Choose a reason for hiding this comment

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

Please test it on GCP, and mentioned it in the description.
Once done LGTM.

@allieychen allieychen merged commit 57ba066 into googlegenomics:master Feb 27, 2019
@allieychen allieychen deleted the vep-bug branch March 7, 2019 19:42
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.

4 participants