[MITgcm-devel] RE: Netcdf - new/different problem?

[Patrick Heimbach]

Hi Laure,

two things:

(1)
As you saw from Jean-Michel's and Ed's mails to the support
list the seg. fault problem seems to be fixed.
So if you don't mind, I'd ask you to try once again
to update to the latest version and see whether
this problem persists:
cvs update -D now -d -P 
or add a redirect if you'd like to see the changes:
cvs update -D now -d -P >! update.log

(2)
It's not clear to me whether your problem below is related 
to the above issue, at least not at first sight.
I have been running the model with MNC over 14400 timesteps
without the problem you described.
I'd like to ask you to try your longer runs with the
latest corrections (hopefully now without seg. fault)
and see whether your problem persists.

Thanks
-Patrick

  > -----Original Message-----
  > From: Laure Zanna [mailto:zanna at fas.harvard.edu] 
  > Sent: Monday, October 11, 2004 5:53 PM
  > To: Patrick Heimbach
  > Subject: Netcdf
  > 
  > 
  > Hi Patrick,
  > 
  > I took the old version of the model from my own backup. If 
  > run the model 
  > for 1000 time steps, I get an output and everything seems fine.
  > I've tried longer runs : 2000, 5000 ... time steps and I 
  > get this after 
  > some time:
  > 
  > NetCDF ERROR: Invalid argument
  > MNC ERROR: writing variable type 'iter' within file 
  > 'mnc_test_20041011_0002/state.000001.nc'
  > STOP ABNORMAL END: package MNC statement executed
  > 
  > Sorry again to bother you all the time with those stupid questions!
  > 
  > -Laure
  > 
  >