Core Minutes 6/19/2012
Core Minutes 6/19/2012DOE Science and Technology Review: (Richard) The review is in progress this week at SLAC. Plenary talks were yesterday, breakout sessions today (Cosmic Frontier session includes Fermi, starting with Rob at 8:30), close-out tomorrow.
ScienceTools: (Jim) There are several new items in Science Tools Development Notes since last week. Among them are
Richard and Jim will be making a trip to LLR to meet with Steve Fegan and Emilia and discuss how they can contribute to ST.
FSSC: (Alex) starting to think about (Mac) Lion. (Tom S.) just got access to a Lion box (someone's desktop) but hasn't yet had a chance to log in. (Heather, Richard) will revisit possibility of acquiring a Lion box (exactly what we don't yet know) at SLAC.
Server migration: (Tom G.) There was a data handling meeting last week at which the migration of 13 nfs servers to a smaller number of wains and replacement of 25 glastlnx machines with a smaller number of considerably more powerful machines (12-cpu rather than 4; 48 Gbytes memory rather than 4) were discussed. There is some concern about the glastlnx replacement — that processes running on the same machine in the new scheme but not in the old could interfere with each other or hog resources.
Pass7: (Tom G.) Trunc64 reprocessing testing is in progress. (Leon) There are a pair of tasks which should generate the same thing, but don't. The first is a two-stage process The second method produces more hits overall, possibly only for large events. The discrepancy could be due to any of
Magnetic field upgrade (new calculation of magnetic equator) is ready to go.
Pass8: (Tracy via email) Nothing big to report on pass 8 anyway, people are chewing on data sets, a good thing.
Core talk:
Upcoming absences (Heather) Michael Kuss is out until July 1st. Email access will be spotty. Tom S. will be in Barcelona next week. (Richard) I disappear to Scotland on Sat for 2 weeks.
DB migration (Heather) Last week mysql-node03 was upgraded to MySQL 5.5 on Monday. On Tuesday the calib* databases and moot mirror db were moved over there and glastCalibDB was shifted to this node. All went smoothly for SCons RM (which was on mysql-node03 prior to this upgrade) and for all jobs needing to access calib.
(Heather) Yesterday glastlnx14 was moved to the IFZ (internet-free zone). Log-ins are only allowed from another node in the SLAC internal network. No problems have been seen with this: log-ins work as expected; RM builds directed to glastlnx14 are working.
The move of various databases to mysql-node01 is still to come.
SeeVogh (Joanne) After various delicate operations on her netbook (upgrade from ubuntu 11.04 to 11.10; remove shards of old video driver) and various suggestions from Vladimir (SeeVogh Support), she now has a fully-functional SeeVogh platform. One key ingredient in the process was the SeeVogh test room which will be available until about mid-July. If you try it in the morning or early afternoon on a week day Vladimir will probably be there to advise you.
(Heather) What should the next step be? Should we try another meeting? Tom may now be the only one left unable to use SeeVogh. (Tom G.) I've a test set up with Wayne Yu over the next couple of days.
It is now looking like there may be money to keep Evo going for another year.
CMT RM — Windows (Heather) Windows builds have been slow. She's cleaning out old builds on the V disk which might help. In addition, the windows-admins mentioned that there was a file server problem around June 8th that may have impacted the ability of the builds to access the external libraries on the V: drive.
GR validation (Heather) Validation of the P8 branch, using event seeding provided by Michael in order to exclude effects of different random sequences, is in progress. There are two stages:
#1 looks good. There are more visible differences in #2 but it may still be ok.
The same comparisons need to be made for L1. There #1 has already been passed. We're working on #2 using the event seeding.
SCons RM (Tom S.) RM is running smoothly. There is a small issue which it would be nice to understand — on Windows only package names are stored in the db with an extra carriage return at the end — but it's not causing any immediate problems. It had been the case that the extra CR was causing a problem for RM web pages: the test output was not being found. But Karen implemented a work-around (searching for a field "like" the package name rather than identically equal to the package name) which took care of it.
GR vc90 (Joanne) No news this week. She's waiting to hear from users.
previous
|
minutes index
|
next
 |
J. Bogart, Last Modified: