EGI Operations Meeting - minutes 23-8-2010 Minutes taken by Ron Around 30 participants: M David, E Imamagic, Rolandas Naujikas, Tiziana Ferrari, Cristina del Cano Novales, Ron Trompet, T. Antoni, Marcin, Luuk, G. Borges, Andres Aeschlimann, Edgars Znots, Renato Santana, Xavier Jeannin, Dejan Lesjak Luka Filipovic, Gabor Roczei, evamvak , sergio diaz montes, Fotis Georgatos florian zrenner, Antun Balaz, Dusan Vudragovic,Vasileios Karavasilis Stuart Purdie, Anders Wnnen, Michael Gronager, Jan Jona Javorsek ---*--- 1. status of gLite release (Mario) - Mario presents status of gLite releases. glite-3.1: * glite_CLUSTER, To be deployed on lcg_CE or CREAMCE, which allows to publish subclusters. * lcg_CE gite-3.2: * glite-APEL. Issue with CentOS. Mario aks for more responsiveness from early adopter teams when they have volunteered to test a new s/w release. Only 1 (out of 6) EA participated in this staged rollout (Emir). * glite_LB, staged rollout is done. Issue with dependency on log4c. Rpm will be in gLite repositories. ---*--- 2. gLite 3.1->glite-3.2 migration questionnaire: - Mario asks for NGIs to fill out the questionnaire and upload it to the document server or send it to Mario. The Deadline is the middle of the week before the Tech Forum. - Tiziana: Is glite_CLUSTER improved mechanism to publish cluster information, will it publish improved information about installed capacity? - Emir: glite_cluster publishes improved cluster information. Has no direct impact on GStat. Emir will check with the accounting people what the impact is. - Mario: glite_cluster will publish more accurate information for heterogeneous clusters. - Emir: Upgrading new glite_APEL on CentOS5 failed due to dependency on jdk. Do you need to install SL5? - Christina: APEL currently needs Sun java. There is either a dependency on openjdk or on Sun java. - Luuk: Dependency or jre is enough. This will install sun-java on SL5 and openjdk on CentOS. - Christina will ask the EMT on this issue. - Mario: Dependency on JRE will take more time. Action on Emir and Mario: follow up on what platform to certify s/w. other tools depend on java, we have to check if open jdk is a good run time environment to try to find a more general solution Several sites are using centos - Luuk: Points to the statement in the installation manual to install openjdk. - Cristina: we have to clarify what is the supported OS, sl5 or not??? - Tiziana: lcg-ce: will it be ported to sl5? No, it will not be ported ---*--- 3. Emir reports on the Infrastructure for Grid Management status. He reports on domain migrations, SAM/Nagios deployment and GOCdb4 migration. ---*--- 4. COD issues (Marcin) - COD collects underperforming sites regarding reliabilility and availability. There is a suspension procedure - Antun (chat): the site suspension procedures (regarding sites below 50% Avail. for 3 months in a row) had been approved by NGI managers on OMB 17.08. - Marcin asks for input on 4 underperforming sites from AP ROC, Ibergrid and ROC Russia. - Marcin will open a ticket for site suspension of UNICAN belonging to Ibergrid. Goncalo does not agree with the fact that the three months periods starts before the procedure is official approved. Goncalo will follow up with UNICAN. No representatives from ROC Russia and AP ROC are present. Tiziana clarified in the chat that the policy of suspending sites after three consecutive months of availability lower than 50% was approved about one year ago, so while Goncalo's concerns is valid in principle, it does not apply in this case, since the policy hasn't been changed since then. ---*--- 5. Actions - Mario will update the action of the glite-3.1 to glite-3.2 migration. Mario will open tickets to NOC managers who have given feedback in this issue. - Tiziana will check the status of CondorG submission to CREAM. While preparing this minutes this action has completed. In the July 17 release of CondorG, two blocking issues were solved (proxy renewal and gridftp client overload). Massimo Sgaravatto already contacted ATLAS representatives to ask for a test campaign. Jamie Shiers will contact Kors Bos to officially appoint a tester. There are no other ATLAS blocking issues preventing the deployment of CREAM. I also posted a request to inspire-sa1 asking people to comment if any other VO has issues concerning the CREAM glite3.2 deployment. No answer so far. - Mario will check the impact of glite-cluster on the installed capacity computed by operational tools such as the gridmap and gstat2 (action to be confirmed by Mario) ---*--- Next will be on September 6th 14:00 CET.