Core Minutes 9/15/2009

• ScienceTools: No one was present to give a report, but you can read it on your own.

• FSSC: (Eric) is looking into support for Fedora 10 [RHEL5 corresponds to Fedora 6; it is rumored that RHEL6, which doesn't exist yet, will correspond to Fedora 11], but has not even managed to build ftools there. However it is possible to use binaries built elsewhere.

  FSSC has a new Mac mini. They plan to install snow leopard on it and then will port to that OS.

• GR RHEL4 Validation: See comparison plots in this pdf file from Anders. Most plots show no discernible difference between RHEL3 and RHEL4. There are a few Cal plots, however, with small differences.

  (Anders) fsw has made a RHEL3 build of 2-1-0 for us. By the time we move to 2-2-0 3 or 4 weeks from now we're expecting that RHEL4 validation will be done, so we haven't asked for a RHEL3 build of that release.

  (Richard) There is a problem on RHEL4 with xrootd pre-load libraries. The knowledgeable person at CERN most able to track this down is just back from vacation so we may see progress soon.

  Background simulations (Heather)summarizes thusly:

  As we working towards validation of the RHEL4 builds of GlastRelease, differences have been highlighted in our background simulations in the system tests when run on RHEL4 versus RHEL3. Meanwhile, this behavior does not appear for single particle runs either for monoenergetic or in the flux runs. Testing has focused on GR v17r34p0 as well as the L1 processing version v15r47p12gr1. The differences in our background runs may be due to changes in the input particle distribution. A quick look at the warnings associated with the packages responsible for producing the background distribution does not seem to reveal a clear smoking gun. While work continues to explain the changes, testing of real data processing has begun to validate GR on RHEL4 for level 1 processing. A summary is available on Confluence.

  Squishing warnings (Heather) There has been some progress. No terribly significant bugs found yet.

• More GR (Leon) is finishing up tracker materials audit with help from Robert.

  (Heather) Now that he is or is about to be back from vacation, we're hoping Tony Waite will make a RHEL4 build of 1-1-3 for us, which should eliminate the warning messages about library mismatch we now see when running on RHEL4.

  ROOT 5.20.0-gl5 (Tom) This version, incorporating a new xrootd client, seems ok. There is no obvious huge improvement in performance, but it does fix a known bug. See plots from Wilko.

  (Richard) We need to encourage Liz to add data sys tests. (Anders) All the elements are there; she just needs to find the time to put them together.

• Documentation: (Chuck) sends the following:

  I've been working with Joanne, Heather, and Tony on updating the "Secure Shell for Unix/Linux Users" page; revised version has been posted to the web.

  I'm also updating the remaining ScienceTools tutorials, and am currently working on the LAT Gamma-Ray Burst Analysis page. Beyond that, I'm handling routine updates as they arise and gathering information on pointlike.

• Disks: (Richard) We're initiating an order for more disk; we'll use up what we've got by about Xmas. We have no plans to order more tape until we have more money to spend.

  (Tom) has made a table of nfs disks and servers, indicating which are critical. Currently about half ot the 13 servers are known to handle critical functions. More nominations are solicited. (Richard) Work on cooling at SLAC is ongoing and will continue through the calendar year. If it gets hot, SCCS will have to turn something off. We have favored status, but still might be affected. If so, SCCS will use this list to help decide what should be turned off.

• Core talk:

  • Tagging gyrations (Joanne) During the spate of tagging last week to eradicate RHEL4 warnings it sometimes happened that in packages with proper SCons-style tags, some files would not get the corresponding CMT-style tag. This could happen when the SCons tag was applied in pieces (necessary for "non-leaf" packages like celestialSources). There are new versions of stag (0.2.1) and GoGui (9.6) available via ftp which avoid the problem by tagging the pieces in a particular order (and only tagging in pieces when absolutely necessary which makes for better performance). They have also been installed for SLAC rhel3 and rhel4 machines.
  • Wrappers (Navid) Status is more or less as it was last week: new wrapper code is working for Linux and is in progress for Windows. It doesn't help that jscript does not behave as documented.
  • RMViewer 1.0 (Heather, Navid) This version, incorporating a tag collector, is available from ftp and is installed on the Terminal Server. Heather is happy with it so far. It's important for other potential users to try it out. Remember it's live! Tag-collecting operations have real effects. But there are plenty of other read-only features to exercise as well. (Navid) anticipates continued development to
    • improve user interface
    • improve performance
    • improve code organanization
    but not to add new features. Those, such as ability to control RM activity, will go in 2.0. (Heather) Are there command-line tools for SCons RM? (Navid) Yes. There are tools in the glastrm directory for deleting old builds (tested) and triggering new ones (not yet realistically tested). There is nothing yet for redoing a build (Heathear) but one can get around this by triggering an essentially identical new build instead.
  • GR and SCons (Joanne) has made only a little bit of progress since last week. After consulting with Tracy she's midway through a reorganization of G4HadronSim
  • ASP and SCons (Joanne) It's slowly chugging along. No obvious roadblocks, but plenty more to do.
  • RSP and SCons (Tom) expects some lessons will be learned from ASP conversion. (Joanne) Yes, to a degree, but RSP depends on both ScienceTools and ASP, which makes for still further complication. We might want to re-evaluate this design.
  • RHEL5 externals (Heather) Kim is working on it; no particular news this week.

previous

minutes index

next

J. Bogart, Last Modified: 01-Jun-2010 15:46:26 -0700