Ibrun does not return exit code?!

[iparask]

Hello Andre,

I noticed the following happening. When I run a MPI program with ibrun on Stampede2 I see the following:

Success:

TACC:  MPI job exited with code: 1
TACC:  Shutdown complete. Exiting.
c455-041[knl](18)$ RETVAL=$?
c455-041[knl](19)$

Failure:

TACC:  MPI job exited with code: 1
TACC:  Shutdown complete. Exiting.
c455-041[knl](22)$ RETVAL=$?
c455-041[knl](23)$

I am not sure if RP is then able to get the correct exit code. Any ideas how to fix this? I will continue looking.

[andre-merzky]

Hmm, interesting, that works ok on stampede_1:

c557-904.stampede(1)$ ibrun -n 1 fail
TACC: Starting up job 8712327
TACC: Setting up parallel environment for MVAPICH2+mpispawn.
ERROR: The -n option to ibrun was set but -o was not.
TACC: MPI job exited with code: 1
c557-904.stampede(2)$ echo $?
1

I think this should be an XSEDE ticket. Would you mind opening one? Thanks!

[iparask]

Wait, I missed doing echo $RETVAL!

[andre-merzky]

Hahaha... :)

[iparask]

My mistake. It works...... Apparently the application I am running had a bug and did not return something upon completion