Minutes for the teraminutes of 22nd January 2007 Present: Mireille, Fred, Henry, Emmanuel, Chiara, Laurent, Yannick (partly), Mathias (by teleconf)

Emmanuel's summary document

MEGACAM PI data
 Barger data has not been processed by QFits because of a problem with datatransfer.
 Datatransfer still doesn't work; Laurent will try to fix it soon as possible (footnote: it seems that this has been fixed thi afternoon). Once this problem has been resolved then the CFHTLS-T04 data will also be transferred to the nodes.
 Processing PI data Megacam requires intervention of Laurent / Fred and data transfer which can add a lot of time to the processing; is there a way to speed things up?
 Need to have someone to follow the processing of PI data at a higher level: Henry has volunteered to do this. We will also take care of the quality control on the "Wide" part of T04.
 Mireille will explain to Mathias how to make the stacks for the 04AF09 A. Robin data in February.
 Emmanuel will try look at the problem of the z-band scamp solution in the 04AF09 z-band data next week.
 Question from Chiara: What shall be the policy concerning the storage of old PI data images? It is suggested that PI "input" data should be conserved for one year (for a possible reprocessing); after that it will be removed from the disks. Long term observing projects might be offered from a longer grace period.
 Laurent has installed W3Perl on FTPIX. A small demo is made during the meeting to show how to trace downloads on this machine.

T03 interface
 (footnote: Laurent has fixed the broken links.)
 Laurent will fix the broken CVS export function (footnote: done).

T04
 Laurent has tested QF on mix3 with the new optimised version of SExtractor; that has been running for one week now. Only one QFITS is running at a time. Laurent may add another QFITS job if needed at a later time.
 Emmanuel still needs to look at the options for flux conservation in SWarp.

Software
 Skymaker 3.1 has been released.
 SExtractor 2.5.1 has a 25% speedup with a 'patched' intel compiler.
 Question from Chiara about the pixel scale in PSFEx check-images: the pixel scale defined in the QFITS PSFEx configuration file is the *same* as the pixel scale for the input images.
 One task for the new engineer could be a program to automatically evaluate the PSF quality for each image, based on the PSFEx output.
 Chiara asks if she is allowed to release her IR processing scripts to P.I. when their data cannot be reduced fast enough at TERAPIX.

Machines
 32 megs now installed in mix7.
 mix4, mix5, mix6 RAIDs have been rebuilt and are now working.
 Pix6 power supply has been replaced. Pix7 has turned unstable and freezes after being up for some time.
 The black 32bit machines (pix5-9) are actually on UPS. Pix1-4 are not, but Fred is waiting for a small UPS provided by Carlos.
 Breaker pb: Fred will recontact Powerware today.
 What is the status of PI data on pix6,7: Mireille will have a look and report back to Laurent.
 Need to order a new power suppply? Only one spare is left.
 Laurent will review of all the PI + calibration data on the machines.
 It seems that the problem from the slowing down of SWarp is because of very long and intensive SMART self-tests; these are now turned off.

Other matters
 Several old dot-project tasks were removed. Laurent will ask Jean-Christophe about the current status of SkyWatcher, and upgrade his dotproject entries.
 Laurent Cambresy in reality doesn't do sky subtraction. Mathias will contact people at IPAC (Spitzer processing) this week to get processing tips. Answer expected for next week.
 New QF has been installed in Naples.
 Gregory Semah has accepted CDD post at TERAPIX; his background is in signal processing. Yannick will be here on Monday 29th to welcome here on his first day.
 His role at TERAPIX will to automate processes at TERAPIX.
 CADC have requested CSV tables for T03
 Emmanuel showed his TERAPIX presentation for next week at La Rochelle and asked for input. A few details were fixed. A beautiful image must be added.
 Fred corrected the link to T01 field map during the meeting.

Wircam
 Deharveng data processed
 Only willot data better than 0.9'' will be conserved, first pass images processed.
 a lot of work still to be done on automatic quality assessment.

Next tera-meeting: to be defined during the week (depending on the availability of TERAPIX members): not before Feb 1st.