[MITgcm-support] MDSREADFIELD problem

[Jean-Michel Campin]

Hi Li,

two things:
1) it's better to use the MITgcm-support list to report this type of 
  problem.
2) the error message is easier to interpret than the problem you had 
last time (Jul 14: Subject: [MITgcm-support] a quick compile question) 
but the reason is the same:
I think you are mixing pieces of code form 2 different versions
and the error you are getting here "ABNORMAL END: S/R MDSREADFIELD is retired"
indicate that in some (old) source code you still have a call to this
MDSREADFIELD subroutine but the routine has been "officially" retired
(with a stop in it).

Thanks,
Jean-Michel

On Wed, Jul 21, 2010 at 03:07:38AM -0700, q li wrote:
> Hi Jean-Michel,
> 
> Sorry for bothering you in this way.
> 
> 
> I had this error message.
> 
> [hiphop at node4 run]$ mpirun -np 1 ./mitgcmuv < /dev/null &
> [1] 30678
> [hiphop at node4 run]$ ABNORMAL END: S/R MDSREADFIELD is retired
> rank 0 in job 20  node4_46608   caused collective abort of all ranks
>   exit status of rank 0: killed by signal 9 
>  
> [1]+  Exit 137                mpirun -np 1 ./mitgcmuv </dev/null
> 
> I saw a post 
> (http://mitgcm.org/pipermail/mitgcm-support/2006-December/004495.html), in which 
> you mentioned it is because the file does not exist. My file is exactly in the 
> same directory.
> 
> I also searched signal 9. It seems to mean "Exit code 9: Ran out of CPU time. "
> 
> I ran my program in another cluster with MITgcm version before Mar. 2009. Now I 
> changed to a new cluster (AMD64, ifort). I have attached my opt file. Is this 
> caused by my code or setting, or caused by my cluster setting? There is almost 
> no error message. I cannot search on internet. Any advise would be appreciated.
> 
> Best regards,
> 
> Li
> 
> 
>