Terameeting for August 19th 2005 Present: Mireille, Fred, Laurent, Yannick

Status of T0002

 The pending problem reported in the previous Teraminute: " -4. There is also the problem that one pre-survey images are actually in the wrong part of the sky, we don't know what this comes from. This image should also be removed from the archive and CADC repository. " turns out to be irrelevant for the release. This image was obtained before June 1st, 2003.

 The "denied access" to cfhtls users to the detailed image description of each stack from the T0002 : Wide and Deep summary page has been fixed. Access was verified by Mireille, Liping and Yannick.

 Links to each Qfits-outs is under way. Laurent will provide a complete list, like: 750309p_V0002 , and Yannick will add these URLs in table T0002 : Wide and Deep summary during the week end.

T0003

 Laurent made several tests of SPICA over the week. SPICA can now run two Qfits-in per node, by running one script one each raid1/2.

 Fred will test this week that all master-files needed are indeed available prior to start running Qfits-in.

 The Qfits-in will run SCAMP in order to provide information on astrometry of each input Megacam image. New SCAMP meta-data will then be produced that have not be included in the ICD. This document must be updated.

 Laurent is going to produce a "ldac repository" where all Qfits-in ldac files will be copied. So in case one node failed, SPICA will get the ldac in the repository. The repository will be created by Fred on mix5 or mix6.

 Fred has not yet implemented a local access to USNO catalogue. Should be done quickly. In order to avoid problem if the node that hosts the local USNO catalogue is down, the script written by Jean-Christophe will be modified to offer the alternative option to go to CDS is case of local access failure.

 Laurent needs skyprobe, .log files and other relevant metadata of PI programs. The request was sent this week to Jean-Charles who answered he will take care of this over the next days.

 urgent: Fred's daemon has to rename failed stack from .spica to .error.

 Laurent is preparing a simulation of T0003 in real conditions, using mix3-5 and pix4-9. It will be done once Fred will have installed Gentoo on pix*. Likely beginning of next week.

 We agreed that for the next Qfits-in we will use CFHTLS data (*L*, *Q98 and *Q99 RunID). The others (*Q97 = SA* Landolt fields) or PI) will be put on SPICA disks later. The PI *K* (Korean), *H* (Hawaiian) or (*T* Taiwainese) data will never be copied on SPICA disks. The other PI data and *Q97* will be put and Qfits-ined once CFHTLS data will be QFits-ed. Even if some later PI QFits-in may change SCAMP for some CFHTLS data, it will be minor changes as compared to the initial setup.

 Fred will finish installing and set up clix2 tonight.

 The T0003 release and next release candidate will release homogeneous catalogues. In contrast with T0002, where Deep and Wide catalogues are different, we must provide easy to handle data where users have not problems of understanding what is what. Mireille, Fred, Henry, Emmanuel and Yannick will meet next Tuesday afternoon to define the catalogue list parameters contain in the release. Also, catalogues names should follow the ICD.

 Laurent's tentative scheduled is adopted.:

 25/08 : stop developpements, configurations frozen. All T0002 data and meta-data saved.
 29/08 : cluster hardware ready (clix2 on, pix5-9 GENTOO, mix6 on). Installation of the new SPICA pipeline done + move to dbterapix done.
 29/08 : Full pipeline tests with all new software releases + new data base + new hardware + new Terapix cluster.
 05/09 : Check SPICA/Qfits-out products + ICD.
 05/09 : bay raid, nodes, data bases cleaned up. Start data transfer from in/out disks to SPSICA disks,
 15/09 : Start Qfits-in processing.

Hardware: (Fred)

 Cosmix4 not arrived yet. Likely on week starting on August 29th.
 mix6 not arrived yet. Likely on Tuesday.