Closed julianpistorius closed 4 months ago
Continuing conversation here, rather than in the closed PR.
This is unexpected. I set up a cluster before issuing the PR and ran this same sample job with no issues.
If mpirun
isn't found, it sounds like the compute node base image didn't actually get set up properly. If it's still around, could you share the entire local_create.log
?
I'm trying again quick to see if I get the same result.
Sorry, I already deleted the instance. :disappointed:
This is unexpected. I set up a cluster before issuing the PR and ran this same sample job with no issues.
Same. So I was also surprised. Hopefully it's a fluke?
Looks like a fluke? I just created 2 clusters back-to-back, one stock-Exosphere-push-button-cluster (so pulling this repo) and one by using my fork, and both of them worked no problem.
I hate to call it a "can't reproduce," but I can't seem to.
Weird. I'll have another go. Thank you for trying.
Yup. Now it works. :man_shrugging: Apologies for the false alarm.
Follow up from #16 and #17.
I just tested #17 by launching a Slurm cluster from Exosphere on Jetstream2, and there seems to be a problem: