Error running a 768 case on Hera

Submitted by biswas on Thu, 08/27/2020 - 19:39
Forums
Medium-range Weather Application

Hi,

I am getting errors while running a C768 case on Hera. I am using 2019082900 (release case). I successfully ran C96 case. 

These are my directories: 

/scratch1/BMC/dtc-hwrf/Mrinal.Biswas/my_ufs_sandbox/cime/scripts/DORIAN_C768_GFSv15p2

The log file showing the errors: /scratch1/BMC/dtc-hwrf/Mrinal.Biswas/DORIAN_C768_GFSv15p2/run/chgres_cube.200825-215831.log. 

I am running the release version of the code. Any help will be really appreciated. 

Thanks in advance

Biswas

 

Permalink

Submitted by turuncu on Thu, 08/27/2020 - 19:45

I have no access to Hera to see the actual error but increasing number processor for CHGRES could solve the problem. Such as, following command will increase it to 144.

./xmlchange task_count=144 --subgroup case.chgres

You might increase more to see what happens.

--ufuk

Permalink

Submitted by biswas on Fri, 08/28/2020 - 18:51

Hi Ufuk and Linlin,

Many thanks for your input. I followed the command you suggested. However, I am getting this error now:

srun: Warning: can't honor --ntasks-per-node set to 36 which doesn't match the requested tasks 240 with the number of requested nodes 4. Ignoring --ntasks-per-node.

srun: error: Unable to create step for job 10895132: More processors requested than permitted

Do I need to change anything else? I am running the UFS for the second time, so pardon my ignorance. 

Thanks

Biswas

Permalink

Submitted by Linlin.Pan on Sun, 08/30/2020 - 05:28

Hi, Biswas,

Another way to fix that is adding one line  "tasks_per_node = 24" after line 771 of file

     ./src/model/FV3/cime/cime_config/buildnml

Here is one successful run on Hera:

hfe01:/scratch1/BMC/dtc-hwrf/Linlin.Pan/test/ufs-mrweather-app-workflow.c768/run

Thanks,

Linlin

  • 75 views

Copyright © 2023. All rights reserved.