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.

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

Use rundir (not absolute path) to select the executable file for integration and parallel tests #2084

Merged
merged 2 commits into from
Dec 21, 2023

Conversation

yantosca
Copy link
Contributor

Name and Institution (Required)

Name: Bob Yantosca
Institution: Harvard + GCST

Confirm you have reviewed the following documentation

Describe the update

This is the companion PR to issue #2083. We have now updated the integration and parallel test execution scripts so that the the run directory (and not the absolute path) is passed to the function that determines the executable file name that gets copied to each run directory.

Expected changes

Previously, if the root path of the integration or parallel test had e.g. "rrtmg", then the "rrtmg" executable would be used for all integration or parallel tests. This update now prevents this from happeing.

This is a zero-diff update w/r/t the GEOS-Chem benchmarks.

Reference(s)

N/A

Related Github Issue(s)

This commit fixes the issue raised in PR #2083 by @lizziel.  Integration
and parallel test execute scripts now use the run directory name and not
the absolute path to determine the executable file name.

test/integration/GCClassic/integrationTestExecute.sh
test/integration/GCHP/integrationTestExecute.sh
test/parallel/GCClassic/parallelTestExecute.sh
- Pass ${runDir} instead of ${runAbsPath} to the exe_name function.
  This will prevent integration test root paths with e.g. "rrtmg"
  from returning an incorrect executable name.

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>
CHANGELOG.md
- Added note that we now use the run directory and not the absolute
  path to determine the executable file to use in integration and
  parallel tests.

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>
@yantosca yantosca self-assigned this Dec 20, 2023
@yantosca yantosca added topic: Benchmarking and Testing Related to CI, integration tests, or scientific benchmarking category: Bug Fix Fixes a previously-reported bug no-diff-to-benchmark This update will not change the results of fullchem benchmark simulations labels Dec 20, 2023
@yantosca
Copy link
Contributor Author

All GEOS-Chem Classic integration tests passed (except for TOMAS)

==============================================================================
GEOS-Chem Classic: Execution Test Results

GCClassic #8774cc3 GEOS-Chem submod update: Reduce timers to essential list only
GEOS-Chem #be203a9d9 Merge PR #2084 (Use rundir to generate executable file name in tests)
HEMCO     #a5d5169 HEMCO 3.7.2 release

Using 24 OpenMP threads
Number of execution tests: 27

Submitted as SLURM job: 13227852
==============================================================================

Summary of test results:
------------------------------------------------------------------------------
Execution tests passed: 25
Execution tests failed: 2
Execution tests not yet completed: 0

Also all GCClassic integration tests were identical to PR #2058, except for:

  • TOMAS (failed)
  • APM (parallelization issue)
  • RRTMG (numerical noise in diagnostic output)

@yantosca
Copy link
Contributor Author

All GCHP integration tests passed:

==============================================================================
GCHP: Execution Test Results

GCHP      #4774e4b GEOS-Chem submod update: Merge PR #2058 (TransportTracers in GEOS)
GEOS-Chem #be203a9d9 Merge PR #2084 (Use rundir to generate executable file name in tests)
HEMCO     #a5d5169 HEMCO 3.7.2 release

Number of execution tests: 5

Submitted as SLURM job: 13227948
==============================================================================

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%  All execution tests passed!  %%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

Furthermore, all integration tests were zero-diff w/r/t 14.2.3.

@yantosca yantosca marked this pull request as ready for review December 21, 2023 15:08
@yantosca yantosca requested a review from lizziel December 21, 2023 16:27
Copy link
Contributor

@lizziel lizziel left a comment

Choose a reason for hiding this comment

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

Looks good!

Copy link
Contributor Author

Thanks @lizziel. I'll merge it in.

@yantosca yantosca merged commit be203a9 into dev/no-diff-to-benchmark Dec 21, 2023
@yantosca yantosca deleted the bugfix/integration-test-exe-name branch December 21, 2023 17:48
@yantosca yantosca added this to the 14.3.0 milestone Jan 8, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
category: Bug Fix Fixes a previously-reported bug no-diff-to-benchmark This update will not change the results of fullchem benchmark simulations topic: Benchmarking and Testing Related to CI, integration tests, or scientific benchmarking
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[FEATURE REQUEST] Pass rundir name not full path when determining executable name in integration tests
2 participants