Minutes of NAF User Committee meeting from 9.3.2011
---------------------------------------------------

Present: Steve Aplin (ILC), Wolfgang Ehrenfeld (ATLAS), Andreas
Gellrich (NAF), Andreas Haupt (IT), Yves Kemp (IT), Kai Leffhalm
(NAF), Shaojun Lu (ILC)

Excused: Harmut Stadie (CMS)

1. Report from chair:

Harmut is away this week and the meeting is chaired by Wolfgang.

For those interested in the continuation of the Terascale Alliance
look at the slides from the Goettingen Grid Project workshop:
http://indico.desy.de/conferenceTimeTable.py?confId=4046#20110228


2. Operations report:

The operations report was given by Kai. See the agenda for the
slides. In the following a few highlights.

A new AFS client will be installed tomorrow, which should improve the
situation.


3. Action Items:

1005-3: email notification for /scratch if full 

This is working. This also includes a warning, if a user has too many
small files. Item closed.

1011-05 request for NX

ATLAS put in a written request right after the last NUC meeting. The
NAF admins are now discuss different strategies.

1012-1: evaluating different stdout/stderr handling for SGE 

The new version of the JSV script found at ~finnern/public/jsv.pl is
implemented in perl to be faster and will localise all output. It was
agreed that the experiments test it in the next two weeks and give
feedback. The experiments should choose appropriate test
users. Wolfgang will inform the VO admins.

1012-2: CERN strategy for batch and AFS

CERN is not doing any special. The default from LSF is to store
stdout/stderr first locally and copy it to AFS at the end of the
job. Item closed.

1101-1 WGS reliability/support/monitoring

New hardware for the login server will be added and a new metric for
balancing is ready for deployment. It was agreed that the VO admins
are informed when it will be put into production and the NAF admins
will monitor this closer for a certain time.

1101-3: Twitter

ATLAS has already a working setup for twitter with the new
authentication. Wolfgang will talk to Kai.

1101-7: limiting number of parallel running jobs 

This is implemented for LHCb. If other experiments are interested they
should write a ticket to naf@desy.de. Item closed.

1101-08 test new JSV

All experiments should test the new version of the JSV.

1102-1 test SL5.6

ATLAS did some basic tests on SL5.6 on the NAF and all is
fine. The other experiments should provide feedback until the
next meeting.

1102-2 CVMFS for the ATLAS software

Yves tested CVMFS on two system and was not satisfied for various
reasons. RPM setup not clean enough. Problems with long standing
mounts. ... ATLAS repeated that it has high priority for ATLAS as it
makes software distribution much easier, solve the CMT problem and is
ATLAS preferred method for 2011.


4. Feedback from the experiments

ATLAS pointed out some deficits in communication which were partly
mentioned in the status report. The communication for work group
server rebooting improved. 

It is not easy to disentangle AFS from other problems like login. The
recent login incident was handle okay and within two days the problem
was diagnosed, the user informed and the work flow of the user fixed.

ILC had nothing to report. Nobody from CMS and LHCb was present.

5. AOB

Next meeting is on the 13th April 2011 at 1pm.