fix PAPI test step hanging on some systems#19372
fix PAPI test step hanging on some systems#19372jfgrimm merged 2 commits intoeasybuilders:developfrom
Conversation
|
Might be worth adding a comment to the ECs mentioning why we're not running |
|
Makes sense I guess. Added |
|
@boegelbot please test @ generoso |
|
@jfgrimm: Request for testing this PR well received on login1 PR test command '
Test results coming soon (I hope)... Details- notification for comment with ID 1845115202 processed Message to humans: this is just bookkeeping information for me, |
|
Test report by @jfgrimm |
|
Test report by @boegelbot |
|
@boegelbot please test @ jsc-zen2 |
jfgrimm
left a comment
There was a problem hiding this comment.
lgtm
as usual, thanks for taking a closer look and raising a PR :)
|
@jfgrimm: Request for testing this PR well received on jsczen2l1.int.jsc-zen2.easybuild-test.cluster PR test command '
Test results coming soon (I hope)... Details- notification for comment with ID 1845153014 processed Message to humans: this is just bookkeeping information for me, |
|
Test report by @boegelbot |
|
Test report by @Flamefire |
|
Going in, thanks @Flamefire! |
|
Test report by @Flamefire |
(created using
eb --new-pr)On some system the
fulltestis hanging with a defunctmake(zombie process) andT+state on various tests containing "attach" in the name. Observed especially on AMD EPYC CPUs.Seemingly
parallel = 1was at some point used to avoid this (without success).This changes to "test" instead of "fulltest" which seems to work and makes sure the ECs are the same removing the now unnecessary "parallel=1" and adding a configopt used only in one 6.0.0.1 EC but not the other where it was seemingly forgotten.
If wanted we can add this configopt also to the 6.0.0 EC but none of the 6.0.0 ECs had it so I just made the 2 6.0.0.1 ones consistent.