Minutes taken by Luca & Nicolo
Volunteers for future minutes: Luca, Andrea Sciaba
Pablo reviews the Tickets:
- Costin can provide an update on new developments in the ALICE monitoring at the next meeting
- HammerCloud items still open
- For SLC6 migration nagios boxes are missing, to be discussed at the next meeting
Marian asks if MyWLCG trends can be stopped, Pablo answers that we don't have a suitable replacement yet.
Cath presents the WLCG accounting reports:
Cath explains that the reports are produced with REBUS plus an Excel file which requires manual editing. The only source of the data is REBUS: sites submit the data to REBUS through the interface, and REBUS also aggregates data from the accounting portal automatically. It is important that the reports are reliable because they are used by Ian Bird and the funding agencies.
Discussion on the use federation vs site names in the reports. Simone remarks that for accounting it needs to be federation names because of pledges, while for availability it needs to be site names to know who should be contacted in case of issues. Cath shows the Tier-2 accounting reports, which are just extracted from the accounting; manual work is only needed for corrections. About the storage, for the accounting it is what sites claim to have in REBUS while installed capacity is taken from BDII. Cath asks to make available all tables needed in the reports with minimal manual intervention. Andrea Valassi comments that we need to understand what the target audience (funding agencies) wants to see.
Marian asks if MyWLCG trends are used in these reports, since sites occasionally report issues. Cath and Pablo answer that they are not used for these reports.
Maria proposes to document the sources of the data in the reports.
Andrea Valassi on WLCG 6 months report
Andrea Valassi presents the half-yearly WLCG availability reports: Andrea shows the bugs encountered and his proposals for improvement, commenting that there should be no manual work in generating the plots/tables.
On slide 4, Marian comments that the issue described is not a bug from the SAM point of view, since the different Tier classification comes from the different experiment topology in the VOFeeds. AndreaV replies that in reports that go to the funding agencies this should be considered a bug. On slide 5, Maarten comments that he is not aware of issues with the ALICE VOFeed that would explain the problem. On slide 10, Marian comments that the use of the federation name "KR-KISTI-GSDC" was agreed with the site. For NL-T1, the availability in the table is the average weighted by HEPSPECs of the SARA and NIKHEF availabilities.
Discussion if the reports and plots should be adjusted to use site vs federation names consistently. Also experiment vs gocdb names; but the experiments cannot go back to gocdb names since the mapping is not 1 to 1. We need to understand who is reading the reports (probably the funding agencies) and ask them if they have any issue.
Discussion will be taken offline and AndreaV will check with Ian Bird before any big change like site vs federation names.
AOB The next meeting will be on June 6th at 14:00, items for discussion:
- nagios box migration
- Costin's proposal for a common monitoring of the CVMFS repositories and proxies. Simone suggests to get the recommendations from the concluded squid monitoring task force