Core Minutes 5/13/2014

• ScienceTools: (Jim) No development news this week. Max has been running gtobssim simulations to use in diagnosing the excess residuals at very low energies (<100MeV) that we are seeing in the binned analysis. I am planning to work on that issue this week along with minimizing the systematics for binned analyses that use energy dispersion for Pass 8.

• FSSC: (Joe) No news this week. (Richard) Data release will happen soon after Tom G. returns from vacation (June 3rd)

• Outages: (Tom G.) The Oracle outage took place as scheduled and without incident. The new Oracle servers are up and running; new versions of Pipeline and Data Catalog are in use. They have also been physically moved to a new location.

  There was another planned outage to physically move xrootd and nfs servers. Somewhat later, maybe connected with the move (or maybe not) the xrootd server wain069 developed problems. These were ultimately fixed, but then one of the new xrootd servers, fermixrd12 crashed. It went up and down a couple times, which made things even worse. Since this is a new server with a lot of free disk space, it would tend to get chosen for new writes. Dell was called. A technician came out yesterday and replaced the motherboard, but left before unix admin had a chance to try it out by booting the machine in its normal configuration. When they did, it crashed again. They have managed to work around it by by using a development machine as server for the bad machine's disks.

  Another one of the new servers, fermixrd11, also crashed, but this was apparently a software issue. It was reset, rebooted and so far has been ok. Redhat has been notified.

  One result of all these problems is that many newly-created data files were not accessible over the weekend.

  (Richard) With the new servers we have entered a different regime. Each one has 350 Tbytes of space (compare to 32 Tbytes on the old wains), hence a failure is likely to be catastrophic. We need new strategies to react to such an event. However the older machines are, well, old, and so will gradually be replaced. Plans are to buy a couple more mega-servers and also an Oracle dev server.

• Core talk:

  • Python (Heather) The new python build (2.7.6) is done! See a description of what it contains and how to build from source (if that's what you need to do) on the Python Confluence page. Note that, as has been the case for all recent python builds, we do not include iPython. If you need it, you'll need to build and install locally. The Python page cited above has instructions.

    She'll make a new GR using this python; it would be good to have an ST release using it as well.

  • WIRED (Heather) is trying to get it to run on fermilnx14. Problems instrinsic to WIRED have been surmounted. She believes remaining issues have to do with job options.

  • RM (Tom S.) Windows ST builds have been turned off [as agreed last week]. Documentation updates have gone into the main SCons RM page or into pages linked to it. Code under ~glastrm in the grits-cpp package is in sync with CVS and builds for all platforms are up to date insofar as they need to be (e.g., didn't rebuild non-Mac binaries for recent changes since they affected only Mac). We will need dedicated machines running rhel5 64-bit and rhel6 64-bit (a couple of each) in order to get Jenkins builds going on those platforms. (Richard) believes the Computer Center will oblige since they want us off rhel5 32-bit.

previous

minutes index

next

J. Bogart, Last Modified: