Core Minutes 12/10/2013

• ScienceTools: (Jim) There are three items:

  • Minos issues reported by Jean are being addressed. Jim has subclassed to better handle error estimates.
  • Added a class to eblAtten for EBL data, using Gilmore tables format.
  • Experienced RM problems on rhel5 32-bit correlated with recent changes to IP addresses. [See Tom's response below].

  See Science Tools Development Notes for details.

• FSSC: (Tom S.) Goddard is closed today because of likely snow. No news.

• Pass8 Reprocessing: (Tom G.) Will start Pass8 L&EO reprocessing today. If all goes well it should complete in a few days.

  There was a dead battery in the oracle03 disk array which could be related to the bad oracle db performance of late. It will be replaced.

• Hardware: (Richard) We have a req. for a new astroserver..should be enough to last us through the 10-year mission. PPA req. for cores (for other experiments, not us) had been on hold but is now released.

• C & A (Richard) Efforts are again being concentrated on MC/data agreement. Philippe is a perfectionist. (Leon) For ToT Philippe is working on a better extrapolation of gain curve. Leon has an unfinished chore to better approximate the actual behavior of ToT as follows:

  Currently, during digitization, when a charged particle crosses a wafer, an energy deposit is chosen by GEANT from a pseudo-landau distribution, which is like a Gaussian with a long tail to high values, and then the energy deposit in each strip is apportioned according to the path length of the particle in that strip.

  But in fact, we should be doing the same landau thing in each of the strips. (We don’t, because the strip stuff is handled after the GEANT part, “by hand”… maybe a bad decision in retrospect.) Anyway, the reason it should matter is that the time-over-threshold essentially picks the maximum energy deposit, so for real data, the ToT tends to get higher than the average for a track that crosses multiple strips. If we handled this correctly in simulation, the ToT would tend to increase with angle, as more strips are crossed by the track.

  I’m trying to see if I can do a better approximation to the truth, but so far, it seems not to make much difference.

  The worst discrepancies are with ACD. No plan yet for that. There is also ongoing work to see the effect of different CAL calibrations. This is all hard to reconcile with the nominal (but soft) "drop dead" date for a new GR.

• Core talk:

  • Mountain Lion (Tom S.) All externals have been built. He is now engaged in building ST manually. Other outstanding issues:

    • Problems with QT have to be resolved in order to natively build RM on the platform.
    • We are still facing the CVS repository impasse. (Richard) We need to escalate this, bring it to the attention of Shirley Gruber.

  • RM rhel5 32-bit problema (Tom S.) Expects this is similar to the other fall-out from the changes to IP addresses of nfs servers, requiring reboots or remounting of disks on those machines which need to access the disks. These builds run on a separate pool of machines. He'll look into it.

previous

minutes index

next

J. Bogart, Last Modified: