Minutes of NAF User Committee meeting from 14.4.2010
----------------------------------------------------

Present: Steve Aplin (ILC), Marcello Barisonzi (ATLAS), Johan Blouw
(LHCb), Wolfgang Ehrenfeld (ATLAS), Andreas Gellrich (NAF), Yves Kemp
(IT), Angela Lucaci-Timoce (ILC), Harmut Stadie (CMS), Alexey Zhelezov
(LHCb)

Excused: Andreas Nowak (CMS)

1. News from the chair:

Every two years the chair and co-chair need to be elected. Reelection
is possible. This is due for the May meeting. Suggestions should be
sent to Wolfgang and Johan, who will check availability. Election will
be done during the May phone meeting, if nobody objects. Otherwise at
the next Face to Face meeting.


2. Status report:

Status report was given by Yves. See the agenda for the report. Below a
few highlights from the discussion are listed:

 - Lustre

A detailed report of the March (27.3. - 1.4.10) Lustre failure was
given.

LMU used Lustre with automatic clean up. Rod Walker from LMU knows
more details.

The Lustre client (1.6) for SL4 will reach end of life June
2010. Lustre support on SL4 nodes might be stopped after upgrade to SL
4.9.

The NAF presented their first ideas about the envisaged Lustre usage
of the experiments. More updates at the next meeting.


 - AFS scratch

AFS scratch under ~/scratch will be automatically created for every
account. ATLAS and CMS migrated old accounts. ILC and LHCb should
check with the NAF admins. The NAF should check, if the default is set
to 2GB per user and if the default can be set by experiment.


 - batch priorities

ATLAS asked for the status of changing users batch priorities within
the experiment. This is needed for for the summer conferences.

 - Twitter

Twitter will be tested for another month or two.

At the next meeting we will discuss current practise for announcements
to naf-announce. The NAF will summarise their current practise and the
experiments should collect their expectations.


3. Action items:

Some updates to the action items where presented in the status
report. Wolfgang will follow up with the ATLAS cmt problem this week.


4. Feedback from the experiments:

The experiments presented their support request from last month.


5. AOB:

Wolfgang asked for an additional set of monitoring plots. The current
batch usage pie charts are only for used CPU time. As long as the
batch system is not nearly full, these numbers can not be used to
estimate the total used share of an experiment or user. This
information is needed. Wolfgang suggested to have another set of pie
charts, which are normalised to total available CPU time.

The next face to face meeting should be either in July or
September. Place will be Hamburg if nobody has objections. Wolfgang
will set up a doodle.