[MITgcm-support] netcdf error (on lightning at NCAR)

Ed Hill ed at eh3.com
Mon Nov 13 18:10:37 EST 2006


On Mon, 13 Nov 2006 14:32:01 -0500 (EST) Stephanie Dutkiewicz
<stephd at ocean.mit.edu> wrote:

> 
> Hi -
> we are having problems running a version of the MITgcm
> on lightning at NCAR, and appears to occur when writing out
> the the netcdf time averages. Error message attached below.
> 
> Has anyone had a similar problem
> or any suggestions on what to try to figure out the problem?
> 
> This is a run with a biogeochemical
> component (ie using ptracers), but the error seems to occur when
> writing out to the tave*nc and/or the ptr_tave*nc.
> The problem occurs some of the time, not all of the time and
> seems not to be completely reproducable (ie will run fine some of
> the time). Possibly a problem
> with some of the nodes on lightning? but we unfortunately
> haven't kept track. I have no problems like this when running
> the same model on the aces cluster at MIT.

Hi Stephanie,

I suspect that a few unfortunate things are going on in this bug
report.  First, it seems that some sort of error is being generated 
within the netCDF library.  This is detected by MNC which then calls 
the routine MNC_HANDLE_ERR(...) which then causes a fault since the
formatted internal write on line 42 of mnc_utils.F is not long enough
to contain the netCDF error string.  How annoying...

Anyway, I'll check in a more robust version of MNC_HANDLE_ERR(...) later
tonight or tomorrow.  After I do so it would be very helpful for you or
someone else to report what the new error message is on Lightning since
that will give us a better idea what the real underlying problem is...

Ed


> 
> -------------------------------------------------------
> ABNORMAL END: package MNC
> ABNORMAL END: package MNC
> ABNORMAL END: package MNC
> ABNORMAL END: package MNC
> PGFIO-F-219/formatted write/internal file/attempt to read/write past
> end of record.
>   In source file mnc_utils.f, at line number 2028
> PGFIO-F-219/formatted write/internal file/attempt to read/write past
> end of record.
>   In source file mnc_utils.f, at line number 2028


-- 
Edward H. Hill III, PhD  |  ed at eh3.com  |  http://eh3.com/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20061113/c0447131/attachment.sig>


More information about the MITgcm-support mailing list