Minutes for the terameeting of November 13, 2007

Present : Chiara, Mireille, Emmanuel, Fred, Gregory (partly), Mathias, Patrick, Yannick

Yannick's summary document

WIRCam (Chiara, Patrick):

 Chiara and Patrick are happy with the very last processing done by Patrick on Walawender' data. Chiara will show the packaging process to Patrick and they will put the whole set on ftp for delivery to the PI.

 There are 5 PI requests in the queue. Patrick will not start a new one at this stage but will keep going on the training and learning Terapix. But Henry could start any of those now!

 Following Emmanuel' questions, Yannick remember and clarify how operation works. All PI requests have to be CC: to Yannick who then may decide whether it can/cannot be queued in the Terapix-PI pipelines and possibly whether its priority may not follow the chronology rule. Likewise, once finished all processed data should be "approved" by Yannick prior to be delivered. In practice, things are more flexible than described in this rule and Yannick relies on Chiara, Mireille or Henry. But usually Mireille asks Yannick prior to send, in particular when the processing seems unusual, very demanding, or when the results do not look good. Also, once in the queue, direct interactions of Chiara and Mireille (or Henry) with the PIs are totally free and of course recommended, with only CC: of emails to Yannick. With Mathias, Henry and Patrick no in charge of operation, the rules will be unchanged, Yannick will still keep control of what can be queued and what can be delivered.

Megacam (Mireille, Mathias):

 Following Mathias's request, we agreed he can use mix7 to work on Annie Robin's data. We expect cosmix5 will arrive within 2-3 weeks.

 Mireille asked whether Jean-Charles/Kanoa sent feedback regarding Terapix's request for the log-file for PI data. Yannick said he sent 2 emails to Jean-Charles and Kanoa but just got acknowledged by Jean-Charles. But Jean-Charles asked Kanoa to proceed.

 Regarding Daniel Rouan's request to keep track the full yy-mm-hh-mm-ss date of SExtractor catalogs attached to single images (no stack), we will propose Daniel to produce an ldac catalog instead of ASCII because it is easy to propagate the date Keywords on top of it. If Daniel is ok with this, we will ask him to make/remind this option at each request he may send to Terapix. Note, however, that the date being in each original FITS image, it is easy for him to recover the relevant date Keywords for each catalog on his own. This information can possibly be written in the FITS header of each Swarped image using the original DATE-OBS, UTC-OBS Keywords.

Software (Emmanuel, Chiara, Fred):

 Chiara reminded the need for the last version of QFITS on the cluster in order to run Condor easily and keep the pipeline as automatic as possible. Fred agreed this is needed; he will check the very last QFITS version available and will then install it on the cluster.

 Fred explained shortly how his scripts that handles data for Condor in the cluster works. Emmanuel is wondering whether the delivery of a tar.gz file to the users is the most practical solution because the user then must gunzip/untar . The delivery on an uncompressed tar file is not an issue. However, Fred is not sure we have room for another solution but tar-ing. He will investigate further.

 Emmanuel asked Fred to put his script on SVN , so that an offiial version is archived and Emmanuel can play/test it immediately (in particular the efficiency of the tar solution).

DataBase (Gregory:

 Gregory wanted to know if the new pipeline should keep the same data organisation as SPICA-I. The answser is no, and we are expecting for propositions from him.

 The meeting with Gregory is Wednesday, November 15, 15:00, room 281.

 There are still not clear decisions of the internal Terapix filename structure. the alternative to initial choice could be tpx-[number]-observatoryname, with number being an incremental release number of an image, instead of a pure incremental file number that increases without regards of the image.

Hardware (Fred):

 Fred got feedback from Agorus and Carri and is now confident we should receive the new nodes by beginning of December.

 The Deskstops should be delivered this week.

 Monnverville's Mac is ordered

 Emmanuel reminds we still have 2660 euros to spend. Fred will order a laptop to Patrick.

 Emmanuel asked why the pix1-4 are still on line. Fred reminded it was a request from Olivier Ricou. Emmanuel will ask Olivier if these nodes are still necessary. If not, they will be removed from the Terapix cluster.

Misc. (all):

 So far , no solution has been found in order to have condor@iap.fr in the teraop mailing list. It would be wise to find a solution to this because we need an alarm when a Condor process is finished.

 Patrick did a presentation of his past technical and scientific activities (thanks Patrick!). In view of this, his own wish and the Terapix needs, we will decide together on which topic he will focus when Patrick will be back in December.

 Yannick sent the options he thought about the Terapix offices. He asked everyone to think about it in order to decide before Nov. 24.

NEXT Tera-meeting: Monday November 19, 10:00, room 281