Minutes for Terameeting of 14th of October 2005

Present: Yannick, Anthony, Jean-Christophe, Laurent, Mireille, Henry, Emmanuel, Chiara

Status of T0003 (Laurent)

 Currently we are processing 600 images each day
 6000 images processed out of a total of 15,000 images (CFHTLS+PI)
 We decided to stop the data transfer of CFHTLS images from CADC to Terapix to the limiting observation date of July 30, 2005. All data after this bound will be part of T0004.
 Should be finished by end of the month or the beginning of next month
 5% of images take 95% of the time to process
 We should be able to start grading the images, probably by groups of 100 in order to avoid the "qfits-grders" duplication/collision in ranking the same images.
 Laurent checked CFHTS grade 4 and 5 images are automatically graded "C" by Qfits-in validation.
 Discussion of error messages logged by the pipeline during the QF processing: this is at clix.iap.fr/admin/problems/. Laurent will include the full date (with local time in hh:mm:ss, 0 One important point: do we know how to re-mount the fibre channel bays correctly if there is a crash?
 It was also noted that the UPS on clix2 is unstable: no decision taken about this
 it was also noted that the installed UPSs on the machines do not automatically trigger a system shutdown if there is a power failure
 from the 3dm pages it seems that these machines mix1 (raid1),pix8(raid2), pix6 (raid2), pix1 have degraded disks
 Laurent proposes to monitor this page
 Emmnauel and Laurent agreed to take care of raid reconstruction. mix1 is the first priority: If another disk failed on mix1 all the data will be lost.
 Laurent points out that the possibility of reprocessing the images from the interface in case of Qfits-in problem no longer works since dbterapix has been droped; this part of the script has to be rewritten.

Status of PI data (Mireille, Jean-Christophe, Henry)

 Albert, Seymour, Tully, Sanders
 QFins Tully data
 JC will ask CADC/CFHT if they can re-process the Sanders et al data which was erronously removed (03BH48 and 04AH46)
 Tully data will processed simultaneously with the Sanders et al data, but on a different node.

Focal plane modelling, .ahead (Henry, Mireille, Emmanuel)

 Mireille has managed to process the Seymour u* data using an i* reference catalogue derived from an image of the same field but obtained one year ago (hence a totally different QRunID) following the method described by Henry and Emmanuel ; all CCDs seems to be good.
 Following Sysiphe, Henry will continue to work on processing the cosmos data... :)

Scamp (Emmanuel)

 Emmanuel will add some new check-plots showing the residuals as a function of position
 He will also try to solve the problems with g* images found by Mireille, which seems to be unrelated to the .ahead problem. Emmanuel need some time... in the meantime don't bother !

Jean-Christophe (Yannick, Jean-Christophe)

 JCM will start a contract at the observatoire de Paris from the 15th of January but already has started to work there; on Mondays and Tuesday he will be at TERAPIX and also for the Tera-meeting of Friday, the rest of the time he will be at CAI, in the Batiment Perrault (office no. yet to be determined)

Webservices (Jean-Christophe, Emmanuel, Yannick)

 Some technical problems still be resolved with VO compatibility; Yannick suggests JC contact Andre Schaff and travels to Strasbourg to fix this
 Deadline for the release of the webservice functionality on the Terapix web pages is the 14th of November

WIRCAM (Chiara)

 Chiara used SCAMP with 2MASS catalogue to get photometric calibration of WIRCAM fields. Seems to work. Henry will supply K-band galaxy counts to Chiara to check .

 Problems with QF: can't use the catalogue generated automatically with Qfits for the astrometric solution; however, it does work if SExtractor is run manually, not sure where it the problem is coming from but it does seem to coming from the weighting of the detections

 4 images out of 10 causes a crash in mefhisto (QF), Chiara is trying to determine the problem and to produce a modified version

 it is not clear how to make the changes because of dependancies between the packages; Jean-Christophe will try to help Chiara to solve this problem

 Henry suggests that the images should be resampled at smaller pixel scales

 17 November - 9 December Chiara will be at CFHT (Emmanuel as well)

Anis Robji (Yannick, Emmanuel)

 He will work on cash remaining in last 12 months of the Astro-wise contract to produce a new visualisation tool, starting from the 1st November; he will be in office 211

ADASS (Emmanuel, Anthony)

 Very little talks on signal processing
 Talk on linux distributions UBUNTU and Fedora core (redhat enterprise); about half of the people use this for the clusters
 FITS standard for representing distortions not agreed on yet

Astro-WISE (Emmanuel)

 15 nov to 18 Astro-wise workshop at Leiden, probably Emmanuel will go, Anis Rojbi as well.

Next TERA-meeting: Friday 21nd October 2005, room 281.