Error using Parpool inside a SC with MCR v98 (R2020a), and SLURM is the job sch. manager

4 visualizzazioni (ultimi 30 giorni)
Hello all,
I was running a compiled standalone app/program that uses the parallel toolbox with MCR v98 (2020a) inside a SC normally, this is, I got the results I wanted. After some other tests, and NOT modifying anything in the compile stand-alone app, I am getting this output error file:
Parallel pool failed to start with the following error.
Error in StackCurrentF/OpenParPool (line 551)
Error in StackCurrentF (line 87)
Caused by:
Error using parallel.internal.pool.InteractiveClient>iThrowWithCause (line 670)
Failed to locate and destroy old interactive jobs.
Error using parallel.Cluster/findJob (line 74)
Unknown type: concurrentconcurrent.
parallel:cluster:PoolCreateFailed
So, no parallel computation. This happens even when I run a small interactive Job with srun that only turns-on the Pool and then wait and then closses it.
What can be the problem?
Any insights, or past experienses with similar problems, might be of great help.
Thank you!

Risposta accettata

R
R il 8 Mag 2024
I previously encountered this error due to the local job storage location being accessed simultaneously by multiple jobs/users, which triggered the issue. I managed to resolve it by implementing the solution provided in the following MATLAB Answer:
  1 Commento
Bello Sebastian
Bello Sebastian il 9 Mag 2024
Modificato: Bello Sebastian il 9 Mag 2024
Thank you for your answer. Yes, that was precisely the solution to the problem. However, one extra step had to be taken: Because Matlab was a module in the server (Super computers), it needed to be loaded first, and then all the neccessary Matlab directories appeared as a hiding directories. After that, the solution of the link worked perfectly.
Note. Even the parallel jobs stores the ones that are Run with both the MCR runtime and the matlab modules pre-installed.

Accedi per commentare.

Più risposte (0)

Categorie

Scopri di più su Third-Party Cluster Configuration in Help Center e File Exchange

Prodotti


Release

R2020a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by