GDB 11th January 2012
The agenda: https://indico.cern.ch/conferenceDisplay.py?confId=155064
Lawrence described the EMI roadmap for the info system, gradual evolution seems to be expected. Didn’t sound to me like the WLCG BDII was going any time soon.
http://indico.cern.ch/getFile.py/access?sessionId=1&resId=0&materialId=0&confId=155064
Marcus made a plea to bring to completion the work of the previous two WLCG information officers (Flavia Donno and Lorenzo Dini). Sounds like the MB may follow up and request sites implement BDII caching, glite-Cluster, staleness checks and ensure correct choice of top level BDII. http://indico.cern.ch/getFile.py/access?sessionId=1&resId=2&materialId=0&confId=155064
Dave Kelsey presented the work of the IPV6 working group. Testbed is underway, Glasgow and Manchester involved to some extent. The group proposes following a dual stack approach. Dave raised The prospect of a WLCG IPV6 day in 2013.
http://indico.cern.ch/getFile.py/access?sessionId=2&resId=1&materialId=0&confId=155064
Doina Cristina Aiftimiei (the EMI release manager) gave a short presentation on EMI middleware Release plans.
http://indico.cern.ch/getFile.py/access?sessionId=6&resId=1&materialId=0&confId=155064
John Gordon asked if SGE sites were sufficiently happy with CREAM that we could now phase out the LCG-CE from the availability calculation for 2012. No complaints but John will email the Tier-2s to check.
http://indico.cern.ch/getFile.py/access?sessionId=3&resId=1&materialId=0&confId=155064
Maarten Litmaath gave a presentation on the impact on WLCG of weaknesses in the SHA-1 Encryption algorithm. This is a potential major headache for WLCG that was not apparent from the rather bland title of “RFC Proxies”. IETF is warning that its not long before SHA-1 (used in certificates and proxies) is cracked. Projects need to move to SHA-2 but this is potentially painful. for example dcache uses jGlobus 1 which only supports sha-1 but if it moves to j-globus 2 then it cannot support sha-1 – also jglobus 2 doesn’t support legacy proxies which wlcg uses. Various other wlcg middleware components need modifications and those within EMI the plan is to do this in EMI-2 in April/May. It appears that many components may need upgrades during data taking this year followed by some kind of big bang in 2013 after the shutdown/conferences.
http://indico.cern.ch/getFile.py/access?sessionId=4&resId=3&materialId=0&confId=155064
Brian Bockelman gave a presentation on using chroot to run SL 5 codes on SL6 systems. Motivation was some sites were pushing to SL6 but experiments didn’t support it. Sites not keen to virtualise SL5 hence use this as a halfway house. Install SL5 directory tree, bind mount a few system directories like proc into it and then chroot into new root dir and you have a poor man’s virtualised system. Still nice trick seems to be that PBS supports an environment variable that job will be chrooted into at the point it is exec’d. Add in cgroup support and you have nice job containment too. Doesn’t Imply we have to do anything ….
http://indico.cern.ch/getFile.py/access?sessionId=5&resId=0&materialId=0&confId=155064
The storage accounting discussion didn’t happen.
