[MITgcm-support] rejected posting

[Jean-Michel Campin]

Hi Helen,

The email you try to send is from:
> From: Helen Pillar <Helen.Pillar at earth.ox.ac.uk>
but he address you registered is:
> helenp at earth.ox.ac.uk
and since those 2 are different, the message is rejected.

If you could try to always send emails to support list from 
the same address (that your register) it would work.
Otherwise, would need to check with admin how to do that
(i.e, multiple incoming addresses).

Thanks,
Jean-Michel

On Thu, Aug 05, 2010 at 04:20:43PM +0100, Helen Pillar wrote:
> Hi Jean-Michel,
> 
> Thanks for your email. This is the reply I received from the support list. Thanks for flagging up my subscription option too. I had thought it would be best to receive the day's postings in one bundle, but I should change this to get new messages as they are posted.
> 
> Best wishes,
> 
> Helen
> 
> 
> Begin forwarded message:
> 
> > From: mitgcm-support-owner at mitgcm.org
> > Date: 5 August 2010 11:43:29 GMT+01:00
> > To: helen.pillar at earth.ox.ac.uk
> > Subject: longer mpi adjoint runs crashing
> > Reply-To: mitgcm-support-owner at mitgcm.org
> > 
> > You are not allowed to post to this mailing list, and your message has
> > been automatically rejected.  If you think that your messages are
> > being rejected in error, contact the mailing list owner at
> > mitgcm-support-owner at mitgcm.org.
> > 
> > 
> > From: Helen Pillar <Helen.Pillar at earth.ox.ac.uk>
> > Date: 5 August 2010 11:43:18 GMT+01:00
> > To: mitgcm-support at mitgcm.org
> > Subject: longer mpi adjoint runs crashing
> > 
> > 
> > Hello,
> > 
> > I'm trying to run a 1x1 global configuration of the MITgcm and its adjoint on the cluster here at Oxford. I have run the forward model in parallel many times without trouble, but am having big problems getting the adjoint to run for more than 89 timesteps before the run is terminated prematurely. 
> > 
> > If nTimesteps <= 89 the run finishes and adjoint data files are output
> > "  "   "   "              > 89 the run terminates prematurely, no ADJ.***.data files are dumped, all STDERR files are empty and STDOUT shows the elliptic solver still churning away.  The error message received is:
> > 
> > 
> > 
> > 
> > 
> > I tried to use gdb to pinpoint the problem but I get a weird complaint:
> > 
> > 
> > 
> > 
> > 
> > 
> > which I think is because I'm not using gdb properly with mpi (I'm trying to get this right now!) If it is helpful, my makefiles are:
> > 
> > 
> > 
> >        
> > 
> > 
> > and my PBS submission script is:
> > 
> > 
> > 
> > I'm trying to run on an intel x86-64 cluster (80 dual-COU quad-core blades, Intel Xeon E5540 processors, 2.53GHz, 24 GiB RAM) using 9x8 cores. The current set up has all the bells and whistles attached (current packages:
> > oceanic
> > gmredi
> > cd_code
> > cal 
> > exf
> > timeave
> > diagnostics
> > rbcs
> > kpp
> > autodiff
> > cost
> > ctrl)
> > (although kpp and gmredi are switched off for adjoint runs using in_admode_set/unset)
> > , so I'll try getting the tracer release tutorial example or a stripped down version of my configuration up and running first. Please let me know if other info would be helpful. Any suggestions on how to fix the problem would be very greatly appreciated!
> > 
> > Thank you,
> > 
> > Helen
> > 
> > 
>