WP4 phone conference, 7/8/2002
- PPARC:
- ZIB: Thomas
- KIP:
- NIKHEF: Martijn, David
- INFN:
- CERN: Maite, Piotr, Olof, German, Sylvain
- Apologies: KIP
Review of actions
- 13: No update
- 17: No update
- 18: No update. Condor not supported for release 2.
- 40: The MR API has been updated after some experience with the
first implementation.
- 58: LCAS testing not finished because of problems with new
Globus installation. Job manager changed in new Globus release (see under
AOB).
- 74: Thomas has been in contact with Andrea from Turin. Andrea
has sent a document describing their software. The implementation cannot
start before end of next week because Andrea is away on holidays.
Regarding the information that needs to be published it seems WP1 may be
satisfied with information from the batch system. This means that the
accounting sensor from the monitoring task is not required. Sylvain has
already provided such sensor but it is not yet in CVS.
- 76: Thomas got an answer from glue saying that the schema will
probably stabilise by next week. Thomas does not see any particular
problem for the implementation.
Progress reports
- Thomas has already sent a report.
- German: will send his this afternoon
- David: not much changed
- Piotr: CDB implementation almost on schedule.
Budapest conference
- Only ~5 people from WP4
will participate. Olof will ask to reduce the scheduled WP4 meeting to
half day. Due to low participation we cannot use the Budapest WP4 meeting
for initiating the planning for 2003. Instead we will have to schedule WP4
workshop at CERN for beginning of December.
- Thomas: propose to have a WP4 integration workshop where all
WP4 software can be installed. Maite suggest to wait until LCFGng and
monitoring deployed. German said that such a “hands-on” should best be
scheduled in connection with the WP4 workshop foreseen for December.
AOB
- Thomas wondered what are the problems with the new Globus?
Maite said she doesn’t know yet. She has been able to install a computing
element but it doesn’t work. The jobs cannot get the output back and then
jumps between the queues until the resource broker considered them as
failed. This week Cal Loomis is back and he has started to investigate the
problem. He has now managed to submit some jobs and it seems that the job
manager has changed with the new release. This means that Maite will have
to restart the integration of the RMS components that works with the
previous Globus job mgr. Maite
will do this after she has finished with the Global Schema definition.
- Maite: LCFGng will be release today or tomorrow. The
configuration files (.h) have to be changed and Enrico will do this when
he is back from holidays in two weeks time. Only then LCFGng will be
widely deployed (realistically in the beginning of September).
Next meeting: 21/8 11:00