Questions on NTASKS, ROOTPE, and submission

[xiangli]

You need to check to see what your mpi library calls the mpi compiler wrappers and use those.
If both mpif90 and mpifort are available I think they are identical or nearly so - either should be fine.

Hi Jim,

I checked that for my case mpif90 should be good.

Now I could successfully run B1850 cases, though there were still some errors after running ./scripts_regression_tests. Do I need to care about them seriously?

Here are the errors I got:





Any suggestions would be appreciated!

Thanks,
Xiang

 

[jedwards]

I think that you have the significant tests working now and can set that aside now. You might want to consider running the ECT test
and comparing to our baseline. CESM2 | Verification

 

[xiangli]

I think that you have the significant tests working now and can set that aside now. You might want to consider running the ECT test
and comparing to our baseline. CESM2 | Verification

Hi Jim,

Yes, I'm trying doing the validation now.

I downloaded input data for UF-CAM-ECT here:



And I tried running this:



There was an error when creating the case. Shall I edit the ensemble.py file?

Thanks,
Xiang

 

[jedwards]

This is another python porting issue. The variable rand_ints should be of type int but it's type float.

 

[xiangli]

This is another python porting issue. The variable rand_ints should be of type int but it's type float.

Hi Jim,

Shall we update this ensemble.py script or just change line 30 to ippt = '{0:03f}'.format(ll)?

Thanks,
Xiang

 

[xiangli]

This is another python porting issue. The variable rand_ints should be of type int but it's type float.

Hi Jim,

Yes, I changed line 30 to ippt = '{0:03f}'.format(ll) and the case could be created. However, the case failed after running for no more than 60 seconds, with the error messages:





Looking forward to your suggestions!

Thanks,
Xiang

 

[jedwards]

Look in the atm.log for clues about the namelist that it couldn't read. Is the filesystem with your run directory available on the compute nodes?
Try rebuilding with DEBUG=TRUE so that you get a traceback.

 

[xiangli]

Look in the atm.log for clues about the namelist that it couldn't read. Is the filesystem with your run directory available on the compute nodes?
Try rebuilding with DEBUG=TRUE so that you get a traceback.

Hi Jim,

Thanks for your suggestions! I set DEBUG=TRUE and reran the case.

Here is the cesm.log:



Here is the atm.log:



Is this related to the line inithist = 'NONE' in user_nl_cam?



Shall we add the following files into user_nl_cam?



Any suggestions would be appreciated!

Thanks,
Xiang

 

[jedwards]

I don't think that's it. It acts like there is a formatting error in the namelist file but I can't tell what it is? can you post your CaseDocs/atm_in file? I'm disappointed that it's not giving source line numbers or sharing the exact problem it's finding.

 

[xiangli]

I don't think that's it. It acts like there is a formatting error in the namelist file but I can't tell what it is? can you post your CaseDocs/atm_in file? I'm disappointed that it's not giving source line numbers or sharing the exact problem it's finding.

Hi Jim,

Please see the atm_in file attached.

Thanks,
Xiang

 

[jedwards]

I don't see what the issue is. You may need to rerun with a debugger or add some print statements to find the exact problem.

 

[xiangli]

I don't see what the issue is. You may need to rerun with a debugger or add some print statements to find the exact problem.

Hi Jim,

Let me try rerun with --debugger. Here is another question I'd like to ask you.

I was trying doing a branch run based on the restart files from the FV2 BHIST 001 run. Before building the case, I copied the user_nl_* files to the case directory and modified the env_run.xml as follows:

Here is where I deposit the restart files:

However, the case failed, with the following messages:

cesm.log:



atm.log:



Are these errors similar to the previous one?

Appreciate your help!

Thanks,
Xiang

 

[xiangli]

I don't see what the issue is. You may need to rerun with a debugger or add some print statements to find the exact problem.

Hi Jim,

This bug was due to incomplete input file. Now it has been solved.

For the ECT test, cases could be successfully submitted:



However, there was an error possibly related to mpi:



cesm.log:



Since other cases such as B1850 and BHIST could run without problem, I am wondering why this error would happen.

Your suggestions would be greatly appreciated!

Thanks,
Xiang