Minutes for terameeting of 30th Jun 2006

Present: Emmanuel, Fred, Chiara, Yannick, Mireille, Henry

Absent: Laurent, Anthony, Anis, Mathias

Main topic: big problems of security this week after the TERAPIX forum and web pages were hacked

 What happened:
 13 juin files were added to the forum/SPIP
 16,17 other files were added to the forum
 23 new attack, new files were added
 26 in the middle of the night all the pages of the forum were removed; Emmanuel was at SF2A, Laurent and Fred were not around

 The forum was restored using a backup; there was no differential backup made; unfortunately some security holes were left
 9am the following day another attack; the attacker gained access to SPIP pages and removed all the webpages
 Forum software upgraded; and new version of forum software came out
 Attacks continue every two hours, but without any success (usually from finding the forum with google
 SPIP remains vulnerable

Problems

 One problem was that we had no access to passwords; a copy of these passwords should be saved somewhere: deadline, today or monday

List of passwords

 Root password (cosmix,clix) [currently the same]
 Forum, SPIP, .dotproject (mix)
 webservice (clix)

 There is a strong need to present TERAPIXians the organisation of all services provided by TERAPIX. Several short presentations will be organised. We expect the first during the July 10 TERAPIX, by Laurent, about the DBs.

 Differential, successive backups should be implemented, which allows us to go back to a given day if there are a problems Deadline: today or tomorrow

 We need to monitor the partitions to see if there are ones which are full Deadline: Monday or Tuesday

 Perhaps we need upgrade SPIP? But it is not certain that the new version of SPIP fixes all the security problems; and upgrades could be difficult. Fred will look at this with Francois Legrand who already experienced upgrades of SPIP. Deadline: End of July

 It is decided that the following software will be periodically upgraded: dotproject, myBB, PHPmyadmin, SPIP, SVN

CFHTLS users meetings

 Yannick proposes 6,7 November as the dates of the CFHTLS users' meeting

Hardware

 "swarp of death" should be tested on mix8,mix9 (Monday) and machines installed Deadline:10 July
 mix9 will be used for WIRCam, mix8 for SPICA
 The machines will have be to stopped one last time to make another change in the electricity supply
 There is still a problem with mix1, which has an instable raid arrays and may crash at any moment; Yannick made rapid copies on the other machines
 It was not possible backup mix1 at CPPN because there was no way to transfer file structures.
 Plan: excess files will be removed from mix1 by all users, and then mix7 will be used for processing PI data; Chiara's data will be copied to mix9 from mix7/mix5 and Emmanuel's SDSS data to EFIGIX Deadline, 14th July. Then mix5 will devoted to SPICA processing

Previous tasks

 New automount: no tests made yet
 UPS: card inserted, no time for tests yet.
 ypserve: cron job is running and verifies if there are too many ypserve processes running; nothing unusual found
 z counts should be given to fred
 Cleaning up the webpages; can keywords be added to web pages to make searching easier
 More generally, we need to organise a meeting focussed and this issue in order to make pages more useful and practical, with much clearer and easier access to most relevant information
 Documentation: only accepted/published refereed papers will be listed, no popsters nor proceedings. Those that ackowledge Megacam will be separated from others that ackowledge TERAPIX for other reasons, in case both are aknowldged, they will be listed in as Megacam.
 New W3 T0003 relaase: rsync to CADC done, need to improve comments in the changelog file.
 D4-u catalogue problem fixed, need to change in ftpix (Fred, could you please confirm?).
 Annie Robin data: processing
 L02/L05 data: processing
 Mireille's disk: installed, working

WIRCAM PI data deep, Chiara

 Chiara has started to reduce data from Simard in J,K: big problem is that the K-data has no dither offsets! This makes the processing of the data considerably more challenging; this means that we have to try to figure out a way to correctly mask all the artifacts
 In any case, the procesing is fine tuned are is no longer a "pipeline-like" process. I cannot be even transposed for the processing of J-band data of the same field.

Other matters, SExtractor

 Some problems with SExtractor with the new version of gcc which broke the calculation of X_IMAGE, Y_IMAGE
 Upgrade SCAMP: on going

Next Meeting : Monday, July 10