[feat] Add support for the LSF scheduler and LLNL's lrun parallel launcher
#2133
Conversation
|
Hello @rngoodner, Thank you for updating! Cheers! There are no PEP8 issues in this Pull Request!Do see the ReFrame Coding Style Guide Comment last updated at 2021-09-13 21:44:17 UTC |
|
Can I test this patch? |
|
Ok to test |
Codecov Report
@@ Coverage Diff @@
## master #2133 +/- ##
==========================================
- Coverage 86.26% 85.97% -0.30%
==========================================
Files 53 54 +1
Lines 9565 9644 +79
==========================================
+ Hits 8251 8291 +40
- Misses 1314 1353 +39
Continue to review full report at Codecov.
|
vkarak
left a comment
vkarak
left a comment
There was a problem hiding this comment.
Thanks a lot @rngoodner for your PR! It looks very good overall. I have only some minor comments regarding the implementation. We would also need the corresponding entries about the added scheduler and parallel launchers in the docs.
|
@AnibalMG could you test if this LSF backend works for your case? |
- remove unneeded 'or 1' - remove unneeded int casts - LrungpuLauncher now inherits from LrunLauncher - replace launcher completed code with suggestion - move comments into if statements - change a list to a tuple - update tests
|
@vkarak I have addressed all of your comments and also updated the docs. I have also re-tested on the LLNL Lassen cluster and everything is still working for my needs. |
vkarak
left a comment
vkarak
left a comment
There was a problem hiding this comment.
Lgtm! I just did a couple of minor style fixes.
vkarak
changed the title
lrun parallel launcher
7 participants
This adds basic support for LSF and lrun. LSF is a scheduler and lrun is a launcher used on the Lawrence Livermore National Laboratory Sierra Systems. I have used this implementation to run tests on the Lassen cluster successfully.
I followed the sge scheduler implementation that worked off the pbs implementation as an example.
Closes #1055