Fixes and updates, bump third party libraries #1619
Conversation
ginkgo-bot
added
the
reg:ci-cd
tcojean
force-pushed
the
fixes_and_check
branch
2 times, most recently
from
7527266
to
db63a9a
Compare
tcojean
force-pushed
the
fixes_and_check
branch
2 times, most recently
from
6d7124f
to
1a107e2
Compare
|
Small note that MPI seems to work with these MacOS runners with small changes, see: https://github.com/ginkgo-project/ginkgo/actions/runs/9290681420/job/25567479064?pr=1619 (only the benchmark tests fail because of a write issue which changes stdout/stderr).
|
Related issue: #1562
See failure: https://gitlab.com/ginkgo-project/ginkgo-public-ci/-/jobs/6963993256#L654 and many others. It seems that HoreKa and NodeJS do not go well together.
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## develop #1619 +/- ##
===========================================
+ Coverage 89.28% 90.03% +0.75%
===========================================
Files 752 758 +6
Lines 60467 61154 +687
===========================================
+ Hits 53985 55058 +1073
+ Misses 6482 6096 -386 ☔ View full report in Codecov by Sentry. |
tcojean
changed the title
|
Note that there is still a failure with the warnings job which is annoying: https://gitlab.com/ginkgo-project/ginkgo-public-ci/-/jobs/7016226162 A workaround for the issue is to add the My intuition is there is something wrong with the docker container, as I cannot seem to reproduce this easily. If anyone has an idea, I would be happy to test it. |
| @@ -29,7 +29,7 @@ jobs: | |||
| config: | |||
| - {version: "latest", name: "cuda-latest/release/shared", "mixed": "ON"} | |||
| name: msvc/${{ matrix.config.name }} (only compile) | |||
| runs-on: [windows-latest] | |||
| runs-on: [windows-2019] | |||
There was a problem hiding this comment.
I guess a more permanent fix would be nice.
There was a problem hiding this comment.
Agreed, but I've seen other pipelines fail with the same problem, e.g. https://github.com/Jimver/cuda-toolkit/actions/runs/8767123552/job/24060026380#step:11:101 so there is something fishy going on with Github's Windows setup in general.
|
Regarding the warnings CI fail, I would prefer if we don't enable the |
| ${add_test_MPI_SIZE} | ||
| "$<TARGET_FILE:${test_target_name}>" | ||
| WORKING_DIRECTORY "$<TARGET_FILE_DIR:ginkgo>") | ||
| if (add_test_MPI_SIZE LESS_EQUAL MPIEXEC_MAX_NUMPROCS) |
There was a problem hiding this comment.
Is it necessary? MPIEXEC_MAX_NUMPROCS is the number of processor on the host. but mpi should be able run more than it, right?
There was a problem hiding this comment.
See this job for example https://github.com/ginkgo-project/ginkgo/actions/runs/9288144680/job/25558834125#step:6:2170
There are other ways to fix this, such as allowing oversubscription, but this could make these already slow tests take a really long time, which I would vote against.
Co-authored-by: Yu-Hsiang Tsai <[email protected]> Co-authored-by: Marcel Koch <[email protected]>
Co-authored-by: Thomas Grützmacher <[email protected]>
|
Any other comments? Otherwise I'll merge this soon. |
|
This PR fixes some build and CI issues, also addresses some problems reported by users, and other release preparation aspects.
In some cases, the commit message has more details.
Closes issues #1481 #1482 #1141 #1562