[MITgcm-devel] sunos, open_copy_data_file, and data.cal

[Chris Hill]

Hi Martin,

  one possibility.
  open_data_copy_file creates a temporary file.
  where that temporary file gets created (i.e. what directory) depends 
on the compiler and the os etc...
  i have seem problems where the location (maybe /tmp) becomes full or 
is unwritable for some reason. i haven't seen that for a long time but 
given the vintage of some of your machines maybe that could be whats 
going on.
  depending on the os there may be an environment variable to change 
where the temp file gets created.

  one thing that doesn't fit is that all experiments use 
open_copy_data_file.

Chris
Martin Losch wrote:
> Hi,
> 
> I run testreport every week on two of our Sun-Machines, both of them  
> have Solaris9, but I run with different compiler versions (forte7 and  
> studio9).
> This week's test (last night) fails for the only two experiments that  
> use data.cal (and the calendar package pkg/cal, lab_sea and  
> global_with_exf). As you can see from, e.g.,
> http://mitgcm.org/testing/results/2005_05/tr_rays1_20050529_0/lab_sea/ 
> run.log
> the model stops when opening a data file (it's data.cal). I had a look  
> in the repository, but I cannot see any changes that have been made to  
> either open_copy_data_file.F nor the verification experiments in  
> question in the last week. Last Sunday things still worked. Before I  
> embark on tedious debugging execises, does anyone have an idea, what  
> might be going on?
> 
> Martin
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel
>