Progress Report, 2 February 2011

Infrastructure Status

• Integration sites: UCSD, FNAL, Purdue, Wisconsin
  • UCSD passing all monitoring.
  • Purdue passes heartbeat, but not redirector tests (needs TFC change).
  • Wisconsin has problematic dCache servers. Mostly passes heartbeat. Fails redirector tests; needs TFC change.
  • FNAL passes neither heartbeat nor redirector tests.
• Production sites: Caltech, Nebraska
  • Nebraska passes all monitoring.
  • Caltech passes heartbeat, but not redirector. Needs TFC change.

Action Item Status

1. Use Monalisa as a monitoring system: the repository / web interface will run at UCSD, got machine for that on Feb 9. Have working session with one of the developers (Ramiro) on Feb 10.
2. Writeup deliverables and milestones: XrootdUscmsTimeline
3. Start maintenance of UCSD xrootd/hdfs system: Done - but there seem to be issues with hadoop at UCSD, very long delays in file access.
4. Start work in converting a physicist's analysis to use Xrootd: Started, but we need to break down this into 2-week-sized chunks. [MT comment: well, I just got code from Ben.]
5. Service monitoring: Using Nebraska instance for now. Have heartbeat and redirector-based monitoring, as described here. We still need random file monitoring, JobRobot monitoring, and alerts.

Other items not from last meeting:

1. Writeup of CMSSW I/O needs: CmsRootIoIssues.
   • Brian (maybe Matevz?) will likely spend some serious time investigating the first 5 issues. Issue 1 is a concern for this project.
2. Checklist for sites: XrootdProductionChecklist.
3. Writeup of development items needed from Xrootd team: CmsRootIoIssues.

Action Items for Next Two Weeks

1. UCSD UAF cluster. [MT comment: I thought Alja will be free after this week to help me on that but we just loaded a bunch of Fireworks stuff on her. And partially on me, too.]
2. Improved service monitoring (missing tests, alerts).
3. Clarify plans for JobRobot with Andrea Sciaba.
4. Fix dcap deadlock issues (Brian).
5. CMSSW TTreeCache management for 4_2_0 (Brian).
6. Upgrade release to 3.0.2; test cmsd throttling from Andy.
7. Update project webpages: remove references to demonstrator, add information about architecture we're working on deploying.
8. Continue Monalisa monitoring investigation.