Teraminutes for the meeting of 9th september 2005 Present: Chiara, Fred, Emmanuel, Jean-Christophe, Laurent

WIRCAM (Chiara): Estimation of background for in WIRCAM: Emmanuel explains why background subtraction is important for infrared data, and what the typical reduction procedure for such data would be. Chiara presented some work concerning how suitable object masks could be generated automatically using sextractor and weightwatcher. More tests are needed, in particular to see how the 'double pass' swarp mode could work.

Status of cluster (Fred, Laurent, JC): All the cluster machines now are running gentoo linux. Still some things not working:

• ganglia
• gexec
• propagation of internat ports
• most seriously 3dm modules don't work on the pix* so we will not get any warnings if there are problems with the disks. Any disk reconstructions will have to be done from the BIOS screen, and not whilst the machines are running. Might be solved by upgrading the cars firmware.

TERAPIX softs datatransfer and snoopix work without any problem on the new cluster.

 Not sure if the stability problems on mix1/mix2 have been resolved, no-one has worked on these machines since the kernel upgrade.

 NOTE: terapix packages should be linked statically to avoid relocation problems.

Software status: Everyone

 QFITS paralellisation (Fred): this is finished but the calculation of the l,b not finished; after that will start QFITS on the pix for the pi data and on the mix for the cfhtls

 SWARP bug fix (Emmanuel): no time yet to fix this bug, hopefully it will be done soon. Emmanuel estimates that it could take one day of work to fix this is.

 dbclient (Laurent): all calls to dbclient have now been removed from the pipeline; Laurent has written perl scripts which populate the spica database. dbclient and dbterapix are not used any more in the production system.

 Swarp header keywords (decided at the last meeting) for the final stacked images can now be generated by using dbspica database (still to be done).

 Skyprobe data (Laurent, Fred): we still don't have an automatic way of grabbing skyprobe data from CADC/CFHT

 CFHT commentary fields is in a different format for PI data and CFHTLS data, which

 PI data is now supplied in a metadata in a format which is not practical to use.

 backup of dbterapix on ftpix functional;

 Fred proposes to make a 'live backup' of dbterapix on ftpix which mirrors dbterapix, in addition to incremental backups

Other matters

 Generation of templates (Henry): Henry will copy all the ldacs from T0002 (which will not be backed up) as soon as possible and begin work on the problem of generating templates for scamp. Fred has explained how his scripts work. Henry appreciates (now) that this is an urgent task, and will have results for the next terameeting.

 mix7 (8 cores - 10 TiB RAID5 disk -- 32 GiB ram) and office machines now ordered.

 PI data Seymour, Albert Juilli (WIRCAM) transfered, Tully in the course of being transferred.

 It was noted that the T0002 "official" release has not yet been made available at CADC. Maybe sometime at the start of October.

 Reformatted T0002 catalogues which Henry prepared this week should be make available to CADC (via Jean-Christophe).