Incident affecting Apel central services completely recovered.
Tuesday 2nd is the last days for sites to upgrade CAs before alarms are raised. FTS version 2.2.3 (SL4) released to production
Attendance
EGEE
Asia Pacific ROC: ShuTing Liao
Canadian ROC: Di Qing
Central Europe ROC: Malgorzata Krakowian
OCC / CERN ROC: Antonio Retico, Nick Thackray
French ROC: Pierre Girard, Osman Aidel
German/Swiss ROC: Angela Poschlad
Italian ROC: Alessandro Paolini
Latin American ROC: Renato Santana
ROC IGALC: Ramon Diacovo
Northern Europe ROC: Ron Trompert
Russian ROC: Lev Shamardin, Victor Edneral
South East Europe ROC: Marios
South West Europe ROC: Christian Neissner, Gonzalo Merino
See report attached to the Agenda.
Apel failures
Comments:
David (SWE): For the sites in SWE the issues are local and highly technical, therefore raised to the developers support. A timeline for the fix is not available yet.
Nick: should the time for solution grow longer than one week is advisable to put the sites in maintenance in order to reduce noise
MPI failures
Cyril commented to the failure observed in the MPI tests. He wasn't looking at the results directly, so he cannot answer on the existence of trends in the failures
James noticed that from discussions held in the SAM --> Nagios transition context an evaluation was made and the sites failing the MPI tests seem now to be reduced to around five/six. It was noticed however that Nagios is not ready yet to test MPI. A timelline for this feature has been set in a week from now, which makes Isabel happy
gLite 3.1 Update 61 introduces, among others, the long awaited FTS version 2.2.3. of particular interest for the sites also the new version of the host certificate of lcg-voms.cern.ch
A new bundle of patches on the 3.1 baseline is being processed these days. Tasks for the early adopters sites will be opened tomorrow.
ROC SWE: Lots of open tickets related to APEL. There are basically small, quite unresponsive sites affected. ROC will provide a document with a timeline to solve the problems of those sites. the ROC will undertake appropriate action
ROC SWE: On behalf of Mario David, once more the discussion on HEPSPEC06 was raised:
As a question of principles, can EGEE/EGI force sites to install non-free software?
Nick
EGEE perhaps can't, EGI may want to, this is a question eventually to be brought up at the EGI transition meeting
Related to this, upcoming sites might want tables of published HEPSPEC06 values related to specific hardware in order to use that reference instead of running the benchmark in their computing back-end.
Nick
this is against the principle itself of of benchmarking and these tables have in the past to contain largely incorrect and sometimes biased values. The fear was expressed that the imposition of a non-free software may become a precedent in the infrastructure. This is understandable but I think that it would be counter-productive for EGI to go along this line. Furthermore this risk is too theoretical to be discussed in this session. It would be like to infer that having a police causes necessarily a country to evolve into a state of police_ _Renato Santana (ROC_LA) reminded that durign one of the recent SA1 coordination meeting it was decided to set a deadline for the sites to run the benchmark. HE asks for this deadline to be clarified as it is difficult for the sites to go through the needed bureaucracy. Nick will investigate and reply
James: the tests results were evaluated byy the ROD team. Issues were raised, some were new (e.g. the need for MPI testing), other corresponding to already tracked bugs. However the overall feedback was positive modulo some physiological fluctuations in the result comparison and there were no showstoppers. So this morning Cyrill flipped the big switch at around 11 AM and now the alarms from the dashboard are generated by Nagios tests. Availability records will be still calculated with SAM tests for the month of March and the resulting figures compared to those generated by Nagios. If they look compatible the report will be fed by Magios already in April. otherwise the exercise will be repeated.
Ticked for sites are being opened by Nagions and associated to the category 'Nagios' in GGUS
In synthesys now we have