[MITgcm-support] diagnostics fill error ...

[Martin Losch]

Hi Ufuk,

did you modify the code in any way or does this happen with “vanilla” code directly download from the git-repository? If you have modified the code, re-do the modifications step by step to see which one leads to the error. The error message implies that some CALL DIAGNOSTICS_FILL() is in the wrong place.

For vanilla code: In a multiprocessor or multithreaded run, something similar has happened to me, where I guessed that somehow the master thread was a little behind. I can’t remember the solution, but you could try to put diagnostics_switch_onoff.F in the list of NOOPTFILES in your build options file.

In general you could try to compile your code with low optimization and see if the problem goes away.

Martin


> On 4. Sep 2018, at 11:28, Ufuk Turuncoglu <ufuk.turuncoglu at itu.edu.tr> wrote:
> 
> Hi,
> 
> I am getting following error from a simulation after couple of time steps
> 
> (PID.TID 0000.0001) *** DIAGNOSTICS_STATUS_ERROR *** from: DIAGNOSTICS_FILL call
> (PID.TID 0000.0001) *** ERROR *** DIAGNOSTICS_FILL: diagName="ETAN    ", expectStatus= 20, pkgStatus= 99
> (PID.TID 0000.0001) *** ERROR *** DIAGNOSTICS_FILL: <== called from the WRONG place, i.e.
> (PID.TID 0000.0001) *** ERROR *** DIAGNOSTICS_FILL: after the last DIAGNOSTICS_WRITE call in DO_THE_MODEL_IO
> 
> in this case, if i deactivate diagnostic package, it runs without any problem. Any suggestion might help at this point.
> 
> Regards,
> 
> --ufuk
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support