Core Minutes 2/12/2013

• ScienceTools: (Jim) No news.

• FSSC: (Alex) We've built for all Linux using the new ROOT. Mac is next: first Snow Leopard, then Mountain Lion.

• Reprocessing (Tom G.) No news.

• Hardware (Tom G.) glastlnx14 (last remaining redhat 4 machine) is about to be shut down. (Heather) User login privileges are already gone.

  (Tom G.) Yemi has now configured two of our new servers with 3 VMs each. Wilko has volunteered to put xroot redirectors on these machines. He wants to see if latency is a problem. Steve Tether has also starting moving services to the new ISOC machines.

  The ftp-glast problems of last week (nfs issues, exacerbated by hyperactive Google bots trying to re-index) are fixed.

• NASA/Ames Field Trip (Richard) visited with Tom G. to investigate use of their super computer center: 150,000 tightly connected cores, tape silo with 120 pBytes, visualization wall, and so forth. The visited was hosted by Jeff Scargle, who has suggested we might used some of those cycles. There are a couple technical issues to be worked out:

  • Their machines all run SUSE, but they have experience providing whatever glibc, etc. version is needed.
  • Firewall: this could impact us especially when it comes to getting files in; getting stuff out is not as difficult.

  Tony would make submission of these jobs look like submission to a batch queue; he has some interest in trying out this kind of thing anyway.

  We also need to see if Fermi project will back this.

  We will need some extra resources from somewhere in order to make the larger simulation runs Johan and others are asking for as well as the upcoming reprocessing.

• Pass 8 (Tracy) Last week's GR tag, 20-08-08, addressed various crashes in Recon output and elsewhere. There is still one outstanding crash. It needs to be fixed, but occurs at a low enough rate that it can be used to create simulation datasets for the Collaboration Meeting.

  This tag is 20% faster than the previous version. Data volume is about twice that of Pass7. Perhaps we can get it down to 1.5. There still is a long list of block items to be addressed after the Collaboration Meeting.

• Core talk:

  • L1 GR tag (Heather) There is a new one which uses the new obf external. Systests are unchanged (they should be since they don't exercise OnboardFilter!).

  • skimmer (Heather) David Chamont is actively looking into the skimmer issue. (Leon) There is nothing intrinsically wrong since he is able to build one that works.

  • SCons on Windows (Joanne) has discovered a variety of small problems interfering with productive use of supersede, especially on Windows. One involved GoGui. There is now a new version, 1.2.1, available for rhel5, rhel6 and Windows in the usual place. Others required changes to gui, containerSettings and SConsFiles. Nothing has been committed yet. She is hopeful it will be ready for beta testers by the end of this week.

  • Mac, RM, and Jenkins (Tom S.) Jenkins died on the mac, but Tony J. restarted it. This is the first such failure and we have no idea what happened or whether it will happen again with any frequency. Tony has turned on additional logging so that, if it does happen again, we'll have better information. But it now seems to be ok so Tom will shift all Mac builds over to use it. Windows will be next.

  • Mac (Richard) There still is some sorting out of color of money to be done before we can purchase the new Macs. (Heather) At least we do now have a place to put themL in the DataFlow lab.

previous

minutes index

next

J. Bogart, Last Modified: