Minutes for Terameeting of Wednesday April 29th 2009

Summary document circulated by henry before the meeting

Youpi develeopements (Mathias M., Yuliana)
 ".ahead" files are now included in the scamp astrometric solution
 however, it is still not possible to have a 'global' .ahead file (such as a wircam.ahead or megacam.ahead) defining the global astrometric solution as well as individual .ahead files for each image indicating whether or not each image is photometric.
 quailtyFITS/scamp integration: for the time being, the scamp plugin searches in the ldac output directory from the user running the plugin; there is no possiblity to use QF results from another user. There should be a possibility to set a custom path.
 Yuliana would like the scamp plugin error messages to be clearer. There is some ambiguity with the error paths currently shown by Youpi.
 The path for the .ldac files are automatically set based on who ran the qualityFITS
 Yuliana will add the bug report.

Sextractor plugin (Greg)
 Sextractor parameter set for T06: We will use the parameter set from T05. We will *not* include the morphological parameters generated by the new version of sextractor.
 The default parameter file output should be t05 parameter file.
 Need to specify the everything on the command line.
 Sextractor history not displayed correctly
 Youpi needs to be modified so that the database model can accept pluins (addin from Greg) .
 It could be useful to have information on each login page concerning processing made by other people in the group.

Youpi processing rights and permissions
 Following a suggestion from Fred, Mathias M. and Greg will investigate assigning unix-like permissions to each processing. This means a granularity at the user / group / others level.
 Deciding and implementing permission handling on Youpi is currently blocking.

 The "Youpix" web server is set up now. All youpi management services will be moved from clix to youpix over the next few days, which should lead to much better reactivity and reduce the load on clix.
 "Mercurial" version control system has been adopted for Youpix, which allows offline commits to happen.

Wircam processing (Patrick, Henry)

 07BF99 downloaded but will only be done in a few weeks..
 Loudieu data finished and sent to the PI.
 Dubus processing finished.
 WUDS processing in progress on mix19. All data has been sliced and first stage qfits completed. There are 4000 images to grade.
 Probably these 4000 images could be graded using the normal methods (i.e., no remote grading). This is around 50% higher than even a large Bouvier run. Unfortunately the remote grading grading system is not yet ready so for the minute, we will try to use the 'normal' methods; after a discussion on Monday morning between Rich, Henry and Patrick this seemed like the fastest (but most
 First thing to do is to check if there are images outside spec.
 07BC23 (Azimlu) data stacked (henry) but unfortunately not released because of photometric offsets in the final stacks. It turns out the reason why the data contains no FOTC values is because the PI chose the wrong observing strategy for J-band which resulting in the pipeline crashing when attempting to process the J-band data. No photometric data was written for any other bands.
 Loic has offered to reprocess the data to add the missing photometric information but for the moment Henry has decided to try to do the calibration using 2MASS. Unfortunately the photometric scaling derived from 2MASS seems to depend on the error cut.
 loic is now reveiwing all phase two submissions at CFHT so hopefully this should reduce the amount of runs executed with a poor observing strategy.
 WIRDS release is still delayed because of names changes.

Development

 Patrick is reading flex manual, which is a way to generate "grammars", and which could be useful for implementing the image calculator.

Hardware
 fcix5 3ware on raid1 there are 80% of drives with drive timeout errors. It seems that these errors do not have any untoward effects on the operation of the raid array.
 however fcix5 raid2 the cache cannot be activated; same problem exists on mix19. Debugging this problem with 3ware customer support is challenging.
 New machines: fred would like to have all the network information for all the machines centralised on clix, rather than having different configs on each machine.
 some tweaking of qualityFITS with the latest version of PSFex
 It should be possible to have the new machines on-line in the next week or two.
 64mb + 2 nodes
 the computer room has been cleaned out.
 should be possible to have the machines stre

processing for T06 (Yuliana)
 all the required data are now on fcix5 [note to Yannick: this is the entire dataset for T06]?
 no decision has yet been made on whether or not the pre-imaging data should be included.
 question from Henry: all data has now been ingested?