[daip] FRING abort, 31DEC10

Eric Greisen egreisen at nrao.edu
Wed Jan 5 15:48:41 EST 2011


Dieter Engels wrote:
> Eric, we have updated the AIPS version 31DEC10 on January 4 and rerun 
> FRING. The error persists.
> 
> We attach the update protocol UPD20110104.122948.DOC
> 
> The area LIBR/INTELCMP contain two libraries from 2010-12-06:
> 
>> ls /usr/local/hssoft/aips/31DEC10/LINUX/LIBR/INTELCMP -ltr
> total 2963
> -rwxr-xr-x 1 stjh342 st  413884 2010-12-06 19:15 libsvml.so*
> -rwxr-xr-x 1 stjh342 st 2612435 2010-12-06 19:15 libimf.so*
> 
> Anything else to test? Shall we provide the data now?
> Regards and all the best for 2011.
> 
> Dieter Engels
> 
> 
> On Thu, 16 Dec 2010, Eric Greisen wrote:
> 
>> Dieter Engels wrote:
>>> Dear colleagues, we recently installed AIPS version 31DEC10 to work with
>>> an EVN dataset. The dataset contains observations of a stellar maser 
>>> using backends with 2048 channels. FRING aborts with ZABORS: signal 6 
>>> received, if we do not diminish the bandwidth (setting BCHAN and 
>>> ECHAN). Sometimes
>>> also "signal 11 received" is communicated.
>>
>> How recently?  I am guessing from the QGET message that you are using 
>> the 32-bit LINUX installation rather than the 64-bit LNX64 
>> installation. The 32-bit one, in binary form, was distributed with the 
>> wrong run-time libraries (we had to switch compilers and did not 
>> update properly). That was fixed quite recently.  Try running a MNJ 
>> ($HOME/do_daily.<hostname>) and see what is updated.  The 
>> LIBR/INTELCMP area is particularly interesting in that regard.  Then 
>> try FRING again.
>>
>> If it still fails, I will need access to your data since FRING is used 
>> around here a good bit with no error reports.  A FITS file (preferably 
>> containing only the calibration data) in a public ftp site is one of 
>> the alternatives for that.

Those are the current libraries so I guess I need access to your data to 
see what FRING is up to...  A public anonymous ftp site - perhaps having 
just the data necessary to cause the failure - will do nicely.

Thanks,

ERic Greisen




More information about the Daip mailing list