Teraminutes for the meeting of Friday July 22nd 2005

Present: Henry, Fred, Jean-Christophe, Emmanuel, Laurent

Absent: Yannick, Chiara, Mireille, Antony

Actions from this meeting

 Jean-Christophe will conduct an evaluation of openPBS/maui and perhaps sun GRIDengine load balancing and job distribution software for the next terameeting.

Status of hardware commands (Fred, Emmanuel):
 bi-opteron with 1GB ram and 8.8Tb ordered (for incoming/outgoing storage)
 four 24" TFT monitors ordered (but note there is a 30 day delay for this hardware
 cosmix3 ordered (bi-dual-core with 8gb ram 1.2TB disk in RAID5).
 two LDLC PCs not ordered yet because a missing reference on their catalog (Abit mother board)
 broken gigabit switch returned to option+ and new a switch ordered (last Monday: should arrive soon)

Meeting about job scheduling systems with Olivier Ricou: (Emmanuel)
 various options presented.
 we still need to decide which of the job scheduling systems is best for our needs.
 Jean-Christophe will conduct an evaluation of openPBS/maui and perhaps sun GRIDengine for the next terameeting.

system upgrades (Fred):
 all 64-bit systems have now have the same configuration and software installed.
 valgrind not installed yet
 clix1 will be migrated to clix2 for the end of next week (before Fred leaves on vacation)
 homogenisation and upgrades is ongoing for libraries and software like cfitsio, ds9...

PI data pending: (Henry)
 More cosmos u* data which Henry will work on (Henry needs to tell Laurent/Fred which data not to delete and where it is).

Status of t003 (Laurent, Fred)

 Laurent is working on pipeline modifications which will avoid running scamp multiple times (scamp bypass) and on using epsilon(deep) when scamping wide data.

 When can we start the QF step? Data still needs to be transferred to the nodes.

Software (Emmanuel, Fred):

 new version of SExtractor released with new object position parameters which should be of comparable accuracy to those produced by IMCAT.

 SExtractor and Swarp now produce FITS-compliant output files.

 documentation and QF optimisation ready by the end of next week.

 related matter: can we run FITSverify on all images produced by the pipeline (this could be a good way to avoid blocking step @cadc)

Other matters::

 Most recent version of all software (for example, ds9) should be installed on all nodes.

 Laurent has tried several times to back up processed data to ftpix but failed in each time.

 some discussion about optimisation of the usage of machines for the T0003 processing.

 Fred has contacted AMD about possible sponsorship of TERAPIX. The response was quite positive and several options were proposed. A representative from AMD will visit TERAPIX in September. Need to check legal points with IAP authorities.

Next Tera-meeting: Friday July 29th