Minutes for the terameeting of Friday April 29th 2005

Present: Mireille, Emmanuel, Fred, Gilles, Henry, Jean-Christophe, Yannick

Meeting started by a nice bottle of Champagne Taitinger kindly offered by Henry for his postion as "French Fonctionnaire for Eternity."

Photometric calibration (Mireille, Henry, Yannick) :

 Mireille has had some problems calculating the offsets between fields based on the galaxy counts, which turns out to be a super-mongo related problem. Mireille is away next week, and Henry will carry out the photometric recalibration for the fields based on offsets calculated by the colour-colour plots and possibly galaxy counts.

 The next question is, how should these offsets be applied? Should we re-scale the images and re-extract the images? Emmanuel suggests multiplying the images and re-extracting the catalogues. We agreed the multiplication is convenient since we do not need to modify the ZP, which will be kept to 30.

Power failure Last night there was a power spike, which cosmix1,cos-1,4, dbterapix, clix2. to fail.

 It seems there is a problem with the UPS on this rack. To be investigated.

 Laurent did not re-start the stacks after the machines were rebooted because he assumed that Fred had done it. Yannick will sent a mail to teralist explaining clearly what the procedure is for restarting processing after power or equipment failures

Other matters:

 Some images were marked "D" in order to prevent them from being transferred to the CADC.

 These images can now be flagged so that they will not be transferred: they can now be given the correct grade.

 Fred and JC will try to fix the problem of the stacks which are missing PNG/PS or PS missing completeness plots.

 For protopsf the fits headers will be corrected by Fred when all the stacks are finished

 FITSverify didn't accept compressed images; it turns out that this comes from a bug in FITSverify, which has now been corrected, and the new version has been installed on the cluster.

 Emmanuel reports on some situations with crowded fields in which the scamp astrometric solution can be fooled by fields which have a small amount of overlap. This could possibly be fixed by re-weighting the astrometric solutions in scamp.

Defectix (Anthony, Gilles):
 Defectix works with the MEFs from CFHT, but does not work with swarped images.

 It could be that this is because defectix does not handle correctly floating point images: Antony is investigating.

 Yannick express his concern about the slow progress on the test and validation of Defectix. Anthony will take care of this on Monday if the problem is not fixed in the meantime.

Hardware (Gilles, Emmanuel)
 Alphas are still not turned off this will be happen soon. The old rack and the alphas should be removed this Friday, after the meeting.
 The ftp server will be ready the 9th of may;
 mix5 should arrive next week
 Fibre channel arrays for storing outgoing data are very expensive; Gilles and Fred will investigate alternative solutions (for example buying NFS-mounted arrays or buying a fiberchannel board).