You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
forrtl: error (78): process killed (SIGTERM)
- Thread starter bashaman
- Start date
-
- Tags
- wrf bug wrf physics wrf.exe
Hi,
Apologies for the delay. Per the wrf.o* file you shared, it looks like you're exceeding your wallclock time.
I was able to find your running directory on Derecho (I hope you don't mind). It looks like since this post, you've resolved that issue (based on the latest wrf.o* file), but it's still stopping. First, in your runwrf.csh script, increase the number of wallclock hours to the max of 12. That probably won't get all the way done, so can you also try increasing the number of processors you're using? Each node on Derecho has 128 processors. You should be able to run up to 384 processors with the size of your domains, and still be okay.
Apologies for the delay. Per the wrf.o* file you shared, it looks like you're exceeding your wallclock time.
Code:
=>> PBS: job killed: walltime 28826 exceeded limit 28800I was able to find your running directory on Derecho (I hope you don't mind). It looks like since this post, you've resolved that issue (based on the latest wrf.o* file), but it's still stopping. First, in your runwrf.csh script, increase the number of wallclock hours to the max of 12. That probably won't get all the way done, so can you also try increasing the number of processors you're using? Each node on Derecho has 128 processors. You should be able to run up to 384 processors with the size of your domains, and still be okay.
Hi, thanks for the suggestion. Since then I have tried a number of iterations increasing the number of processors, but my simulation breaks down in approximately the same general time in the simulation and I have not been running into any more wallclock issues.Hi,
Apologies for the delay. Per the wrf.o* file you shared, it looks like you're exceeding your wallclock time.
Code:=>> PBS: job killed: walltime 28826 exceeded limit 28800
I was able to find your running directory on Derecho (I hope you don't mind). It looks like since this post, you've resolved that issue (based on the latest wrf.o* file), but it's still stopping. First, in your runwrf.csh script, increase the number of wallclock hours to the max of 12. That probably won't get all the way done, so can you also try increasing the number of processors you're using? Each node on Derecho has 128 processors. You should be able to run up to 384 processors with the size of your domains, and still be okay.
Hi,
I just ran a test on Derecho, using your input and namelist. The difference is that I used V4.5.2 to run it (as it's the latest version). Since your simulation seems to be stopping within the first day, I ran a 30 hour simulation and it completed. For my case, I used the following settings in my batch script:
#PBS -l select=3:ncpus=128:mpiprocs=128
If you'd like to take a look, the directory is /glade/derecho/scratch/kkeene/bashaman/wrfv4.5.2/test/em_real. Perhaps you could try with V4.5.2, as well, and see if that makes any difference?
I just ran a test on Derecho, using your input and namelist. The difference is that I used V4.5.2 to run it (as it's the latest version). Since your simulation seems to be stopping within the first day, I ran a 30 hour simulation and it completed. For my case, I used the following settings in my batch script:
#PBS -l select=3:ncpus=128:mpiprocs=128
If you'd like to take a look, the directory is /glade/derecho/scratch/kkeene/bashaman/wrfv4.5.2/test/em_real. Perhaps you could try with V4.5.2, as well, and see if that makes any difference?