Opened 3 months ago

Closed 3 weeks ago

#3433 closed help (answered)

mpiexec example

Reported by: luciana Owned by: um_support
Component: Other Keywords: mpiexec
Cc: Platform: JASMIN
UM Version:

Description

Hello.

I'm trying to run a job directly with mpiexec on Jasmin and I think I'm missing some steps. Can you clarify them for me, please?

I get two types of error: "Cannot open shared object file" and "All nodes which are allocated for this job are already filled."

The test is in /home/users/lucy/ESDM-test-script and the file readme has the instructions that I'm using to run it and the final outcome with the error messages.

Kind regards.

Luciana.

Change History (2)

comment:1 Changed 3 months ago by grenville

Luciana

I don't think the unable to open mca_ess_lsf: libbat.so: is a problem - it's referring to lsf.

It's unusual to run mpi jobs anywhere other than on Lotus, so I'm not surprised there is resource issued.

You could try running with 1 ensemble member on 2 processors (say) and xios on 2 processors and use the "test" queue (see https://help.jasmin.ac.uk/article/4881-lotus-queues).
{I'm not sure if the test queue supports mpi jobs - I am led to believe it does}

Grenville

comment:2 Changed 3 weeks ago by ros

  • Component changed from UM Model to Other
  • Keywords mpiexec added
  • Platform set to JASMIN
  • Resolution set to answered
  • Status changed from new to closed

Closed due to lack of activity.

Note: See TracTickets for help on using tickets.