|Main||EGI.eu operations services||Support||Documentation||Tools||Activities||Performance||Technology||Catch-all Services||Resource Allocation||Security|
Minutes – Grid Operations Meeting – 14 February 2011:
Alessandro Paolini, Anders Wäänänen, Andres Aeschlimann, Dimitris Zilaskos, Drazen Zujovic, Foued Jrad, Henrik Thostrup Jensen, Jan Astalos, John Walsh Luuk Uljee, Małgorzata Krakowian, Mario David, Mathilde Romberg, Michaela Lechner, Peter Solagna, Rebecca Breu, Ron Trompert, Stuart Purdie, Tiziana Ferrari, Tomas Kouba, Tore Mauset , Torsten Antoni, Ulf Tigerstedt, Vasileios Karavasilis, Vera Hansper, Vladimir Slavnic, Wen Mei
1 - Information, MW releases and Staged rollout (Mario)
1.1 – Status of upcoming releases:
- Upcoming soon (possibly this week):
In Certification FTS/A/M 2.2.5
- In staged rollout:
LFC no EA DPM 1 EA has found problems, a memory leak does anyone want to try it also
- Ready for release - staged rollout OK:
lcg-CE 3.1.46 glite-CLUSTER 3.1.4
- glite 3.2:
- Upcoming soon (possibly this week):
Ready for Certification LB 2.1.20 Ready for Certification UI 3.2.9 In Certification WN 3.2.10 In Certification FTS/A/M 2.2.5
- In staged rollout:
glite-CLUSTER 3.2.2 (No Early Adopter)
We will have to evaluate if the LFC and DPM be released into production anyway, since it's the last release of those components in glite 3.1, and the current production release "suffers" from the same problem of memory leak.
Tiziana: ask the NGI representatives if they know who still uses LFC in glite 3.1.
Mario: has list of sites with LFC in glite 3.1, will send a mail directly to these sites and to NOC managers.
Tiziana: stresses the point of near end of support, and the need to evaluate if those site will go directly to EMI or to glite 3.2. By default there will be backport of new features in EMI 1.0 to glite 3.2.
Mario: notifies the upcoming of UI, WN and VOBOX to staged rollout soon. Mentions the importance to gather the feedback for the glite-CLUSTER. The production CREAM does not authomatically configures the CLUSTER, but this can be enabled manually in CREAM.
Michaela: Some EA not yet in the list, for example Oslo.
- Gathering of requirements for EMI 2.0 - deadline was 10 February (Peter Solagna)
Peter: going to discuss and prioritize the tickets in tomorrow OMB, there was an issue with the RT fields to prioritize them, but can now be done, and NOC managers should do it now.
- The staged rollout wiki Staged-Rollout has been updated, and Staged-rollout-procedures contains information about the procedures/steps that EAs have to follow.
- New: management interface of Early Adopters and components:
https://www.egi.eu/earlyAdopters/ being updated (the wiki table of EAs and components will be deprecated in the near future). Tightly integrated with EGI SSO, creation of new EA teams in the portal will trigger the creation of corresponding SSO groups, that will have permission for the RT. New EA will only be added in this portal, not in the wiki anymore.
2 - Operational issues (all):
- Batch system support/integration into the MW and in EMI in particular, all on Best Effort:
Torque: Nikhef SGE: Cesga, LIP, (Imperial ?) - waiting for more information from EMI (Cream in particular) specially for the GLUE2 Condor: PIC? LSF: CERN?, Padova?
Stuart Purdie I'll clarify with Imperial on them continuing with SGE integration.
Tiziana: from the discussions in GDB and others, this is an open point in EMI on how best to support.
- Major changes in procedures, tools, repositories, affecting sites should be announced well in advance (2 weeks or so):
The example of the latest lcg-CA release.
- Publishing site information in BDII (Tiziana). Proposal to phase out the use of:
- GlueSiteOtherInfo: EGEE_ROC --> to be ALWAYS replaced by GlueSiteOtherInfo: EGI_NGI (regardless of the nature of the Operations Centre: ex EGEE ROC or EGI NGI
- GlueSiteOtherInfo: GRID=EGEE --> where EGEE should be replaced by EGI (note that this is attribute is a list, and other regional GRID flags can be added to EGI)
- sites still using these values are available from gstat:
- MAN01_How_to_publish_Site_Information (recently migrated from GOC WIKI)
Tiziana: asks if we still need this flags, or if all NGIs can change it to the new EGI structure. NGIs like Russia or Italy would have to re-flag this value across all their sites. Gstat will automatically get the new labels. Important cluster the sites under the respective NGI in tools like gstat.
Alessandro for Italy: does not yet have a plan, but if necessary they can start the migration.
Vera Hansper I'd welcome such a change - it simplifies the labels
Vera Hansper Alessandro - I think its more a labelling matter
Stuart Purdie UKI is still a ROC, in process of splitting up. I think we'd prefer not to re-label shortly before splitting into NGI's proper. I'll dig out the expected timeline for that tomorrow, and get back to Tiziana.
Tiziana Ferrari Hi Stuart I think that the UKI situation is indeed peculiar, since it's close to split into two ngis, relabeling of sites at this point in time is useless
Tiziana Ferrari UKI is the only corner case, as I'm not aware of other EGEE ROCs that will split in the future.
Mario: raises point that the Asia Pacific, Canada and Latin America, that are still labeled as ROCs.
Tiziana: for Latin America, already contacted and will follow procedures decided form EGI.
- gLite 3.2 support schedule (pag 6): Full support for 18 months and Security only for another 6 months. This applies to previous mw distributions as well:
- e.g. gLite 3.2 and arc 0.8.x would be fully supported until October 2011, with security updates until April 2012
Tiziana: asks for ops for feedback on the best way to proceed, as to go directly to EMI 1.0 as fast as possible, or to stay in glite 3.2 baring in mind that standard updates will end in October 2011, and security updates end April 2012. Also for sites still in glite 3.1, should try to go directly to EMI 1.0 where available. Staged rollout will take time for the major release of EMI. Organize March OMB with the important point of the upcoming EMI release. NGIs to prioritize the components that should go first into the software release procedure.
Mario: comment on the major work when EMI 1.0 is released. Not all components may be ready for EMI 1.0, we will need to know which components will have major changes, evaluate impact in terms of components, for example cream or dpm. Possibility to start before the release due date.
- ARGUS status update: http://indico.cern.ch/materialDisplay.py?contribId=1&sessionId=0&materialId=slides&confId=106641
3 - Operational Tools (Emir)
Mario: a new SAM/Nagios probes will be released this week, to be inline with the new CA release of last week.
4 - COD issues (Malgorzata, Luuk)
5 - AOB
Next meeting: Monday 28 February 2011, 14h00 Amsterdam time