Core Minutes 7/19/2011
Core Minutes 7/19/2011Travelers return: Heather enjoyed her trip to Ireland. Featured attractions were good weather and beer. Richard is just back from BeXRB2011 in Valencia, which turned out to be a good meeting in an attractive city.
ScienceTools: (Jim) There is one new item in Science Tools Development Notes: a new Likelihood tag which copies GTI extensions to the output. A couple more improvements are in the pipeline, one being to implement requested changes in the gtselect interface.
Pass7 readiness: (Jim) is updating ASP to use Pass7 and is also upgrading to rhel5-64 bit.
(Tom S.) FSSC is almost ready for photon data ingestion. So far so good; still need to complete testing. They're on track for ingesting extended events by the end of the month.
(Tom G.) The Astro Server has also been upgraded to handle Pass7 photon data and should be ready for extended data in another 2-3 days.
(Tom G.) Expect two more reprocessing catch-up runs, the second after the switch-over expected on Aug. 1
Pass7 reprocessing: (Tom G.) Other than the scale of resources required [see below] this task will not differ significantly from other reprocessing tasks. He needs to know which GR tag to use, job options, and requested output data products.
Storage: (Richard) We need a Pbyte of tape for archiving, to make room for Pass7 reprocessing output. It costs $120K. Meanwhile, LCLS is getting so much data that they will fill up the tape silo in 2 months. The plan is to upgrade to new drives which write 5-Tbyte tapes. They will still be able to read the old 1-TByte tapes but will not be able to write them. 5-TByte tapes cost half as much per TByte as the 1-Tbyte ones. Unfortunately, this upgrade will not take place quite soon enough for us, so we will have to go with the old ones. See more on this topic in SLAC Forum.
Of the 10 new file servers we ordered, one has been installed. Others might have arrived.
Documentation: (Richard) No news, alas.
Pass8: (Tracy) For a nice summary of the Santa Cruz meeting, see Leon's review given at this week's C & A meeting. The latest GR tag still has some rough spots; aiming for another tag around the end of the week. However, there are already signs of "new science": previously-discarded events which can be usefully reconstructed with Pass8.
Core talk:
More GR (Heather) Joanne made a new tag on the Pass7 branch, v17r35p23, identical to the previous tag except for the astro package. (Joanne) The new astro tag incorporates one of the fixes - the only one made to a GR package - which addressed the gtdiffrsp problem seen last month.
(Leon) Problems have been reported using Fred and WIRED with newer GR tags. A work-around has been found for the first one. Leon — not the original reporter — confirms that WIRED is not working on Linux.
vc90 executables (Tom S.) tracked down the reason for link failures: a mislaid linker option. Now that it has been re-instated, this kind of failure has gone away (so we can concentrate on various other kinds).
GRBAnalysis-scons (Tom S.) RM is now building this new container for all Linux platforms: Snow leopard is coming soon.
Externals path structure (Joanne, Heather) After the last spate of problems and complaints by Linux users stemming from the compiler component in external paths, we were motivated to investigate simplifying these paths. It will be tedious, but should not be especially disruptive. The one strong proponent (Toby) of the current arrangement probably doesn't care about it any more. (Jim) is firmly in favor of the change. Will there be a tool to create the new-style externals from the old which could be used by remote users? (Joanne) There could be. Other parts of the work, principally many additions to allExternals.scons, will have to be done by hand, but, since the changes are to files in CVS, only once. Once the new externals are in use by RM, RM will make tarballs which will be available to the Installer.
Only new tags will use the new arrangement and there is no requirement to change over all externals at once. Probably we will start with just those used by ScienceTools since ScienceTools developers are the ones most inconvenienced by the current arrangement.
Announcements
SCons meeting Tomorrow (Wednesday), 10:30 AM PDT
previous
|
minutes index
|
next
 |
J. Bogart, Last Modified: