Minutes of the
T2UK phone meeting 14/03/07
Present:
Pedro, John, Dmitry, Erkcan, Jiri, Gordon, Monika,
Paul, Simon, Fred, Barry, Nikos.
Apologies:
Thorsten
News
Fred: UK payment of
CtC has been received by ATLAS.
Robins and ROS software
Barry+Fred: The boards with problems are back to
CemGraft
and will be x-rayed again, but the timescale of any
outcome and feedback from them is unclear. At least the
good news is that this is not a wide-spread problem. Many
more Robins have been used since the last meeting (when
this problem was reported) and only one had showed a problem.
The production of an extra 67 boards will start soon in
Germany.
Gordon: The phase-2 commissioning run has involved LAr/Tile and
a few Muon chambers. Generally successful, but problems showed
up and got fixed. In particular, currupted data gave rise to
many error messages and this caused crushes, since ERS was not
completely thread safe. Fixes went in (also from the detectors
to avoid sending corrupted data) and no crushes were seen since
then. Currently, a discussion is on about how to deal with
requests for events that are not in the Robins. The most
favourable solution currently seems to be to re-queue these
requests.
Gordon+Jiri+Erkcan+etal: The next technical run will take place
the week of 19th March and will focus on HLT integration. Jiri
and Gordon will be heavily involved in this, Erkcan, Dmitry and
John will take HLT shifts.
Farms
Fred: 1st rack of Dells installed. 3 more racks are there and
will be ready to use soon, once some power distribution issues
are sorted out. Finally one reack of woodcrest machines should
be available by mid-april. A 2.3GHz quad-core machine is ready
for delivery. It will go to RAL next week and soon after to CERN.
No more CPUs till the end of the year.
Networks
Fred: Additional equipment is being ordered for the 2008 farms.
HLT Software
Dmitry+Julie: Migration to configurables of
T2UK packages
Erkcan et al: problem with inefficiency in electrons traced down
to differences between offline & L2 spacepoints. Using the offline
space points, we recover all the efficiency losses. The problem
seems to be in the code forming the SPs from clusters and evaluating
their global position.
ACTION: Erkcan, Dmitry and Nikos to get to the bottom of this.
Dmitry: timing looks ok in 12.0.6 - to redo with the new BS converters.
Jiri: produced dedicated release for the technical run.
Paul+Simon: continuing work on
TriggerTool. Awaiting feedback on the usability
of the tool from next week's Technical run.
--
NikosKonstantinidis - 28 Mar 2007