Teraminutes for the meeting of Tuesday 27th of September

Present: Mireille, Chira, Fred, Jean-Christophe, Henry, Yannick, Laurent, Emmanuel

Summary, Yannick's proposition starting T0003

 On Friday Sept. 30 afternoon the RAID disks will be erased and "datatranfer" will be started to move data to the nodes; this will take around three days;

 After this, QualityFITS will be started (probably remotely) and we will be attempt to monitor the progress of the stacks remotely. Fred will be in charge of Qfits-in remote control, Mireille and Yannick will check the Qfits-in products and will comment to teralist.

Status of the cluster for T0003 (Fred, everyone)

 Laurent has problems copying the files between the nodes using 'rcp'; after a certain number of files the copy times out.

 Fred suggests for the moment that Laurent should modify his scripts to use scp even though this is less efficient. If scp failed, then Laurent will move to wget. NDLR: scp works !

 Laurent would like launch 'burn-in' stacks from T0002, on the machines to see how stable they are; for mix5 and mix6 which didn't exist at the time of the T0002.

 Laurent complained several important emails he sent to teralist got no feedback. When important emails are sent, please make it clear and send the email several times.

 The T0003 cluster configuration is still under discussion. For the moment we have: mix1 2.6.11-gentoo-r7 mix2 2.6.13.2 -reboot-> 2.6.7-gentoo-r11 mix3 2.6.9-gentoo-r9 mix4 2.6.9-gentoo-r9 mix5 ?... mix6 2.6.11-gentoo-r7 ftpix 2.6.13.2 -reboot-> 2.6.11-gentoo-r7 fcix 2.6.13.2 -reboot-> 2.6.11-gentoo-r7 dbterapix 2.6.13.2 -reboot-> 2.6.9-gentoo-r14 clix 2.6.9-gentoo-r14

 Yannick's proposition is that the data transfer should be started on Friday so that qualityFITS can be started on the nodes while people are away from TERAPIX.

ahead files, Henry

 Has not advanced much, because of other commitments; will be able to devote more time from Friday onwards

 Suggestion from Emmanuel; start with a generic scamp file .ahead -- MOSAIC_TYPE LOOSE; then with the FIX_FOCAL PLANE by image and then QrunID. 05AH18 (70 files) is a good example of data set which have many initial CCDs off.

 Chiara will try this method on the Dave Saunders COSMOS-u data (05AH18), which has many of outlying CCDs. Mireille emphasises this is an important step because even "A/B" grade images have small offsets. These small offets may be responsbible of the final position filter-to-filter residuals discovered by Willot in some Wide stacks.

Status of hardware and software revisions

 Fred has updated the bios on mix5 and the machine seems to be less unstable. He will check mix6 has the right bios version

 mix6, mix5 have kernels more recent than the other machines (see section above)
 Yannick proposes that we keep this configuration for the time being.
 Machines appear to be stable over the last few days, although no really heavy processing has been done yet; yannick proposes that we keep this config for the time being.

ADASS (Fred)

 Jean-Christophe will print his poster and Anthony's at TERAPIX in the next few days
 Emmanuel will present scamp

 Summary of abscences:

- Fred : 1-9 Oct, - Yannick : 2-9 Oct, - JC : 1-6 Oct, - Emmanuel: 1-5 Oct, - Henry : 5-10 Oct, - Laurent : 30 Sept. - 9 Oct.

Next Tera-meeting: on the return from ADASS, Friday Oct 7th