Production redirector will be upgraded to Xrootd 3.1 this week.
Vanderbilt fallback in place
UCSD issues - corrupted CA certs. We now monitor the CERN CA
Caltech issue - out-of-date GUMS server, and Brian's CERN certificate was "new" to the CMS VO. Works now.
Very quiet week so far in overflows, no abnormal number of failures or CPU efficiency issues.
Purdue - disabled for overflow, but used for regular jobs. Will run a CRAB job that "purposely" overflows at Purdue at scale to simulate the conditions where we saw failure. Will start at Purdue-Hansen.
Igor will show Matevz how to do this.
James has handed over the fallback JR to Hammercloud people.
++ Xrootd S/W
Xrootd 3.1 is released. Go get your sites upgraded. This will be the "baseline" release for all sites.
3.1 client is getting worked into CMSSW. Some minor hiccups in ROOT integration.
Matevz working on Xrootd 3.1 and Hadoop 0.19 integration with .
All this software is working its way to an OSG release near you.
++ Roundtable
Igor: What's the usage of the daily operations reports? Folks seem to find it very useful. We'd like to see it hooked up via cron for the next week or so, then we'll convert it to a Gratia-based report.
Matevz: Alja developed a new webpage view for the monitoring - a table of all active files. Will be put in the production soon. Working with the CERN Popularity Service folks to get them xrootd data.
Dan: Still finishing up the Condor defrag work. Talked to Doug Thain about the CVMFS/Parrot work; Doug would be happy to provide consulting. No progress on read-ahead buffer for xrootd-hdfs.
Brian: Talked to CMSSW folks about the FJR changes we'd like.
Ken: Can we think about presenting to the USCMS T2 meeting? Decision: let's just go ahead and do that.