NUC 09.02.2011

Hartmut Stadie, Steve Aplin, Yves Kemp, Wolfgang Ehrenfeld, Johan Blouw

network
- some network problems (10 GE link HH-Zn down for a day; 1 core routen down)
- new firewall with improved performance
- 10 GE infrastructure at HH one HW fix needed (on Friday)
batch system
- new SGE shares will be effective next week
AFS
- each experiment has two dedicated file servers
- have seen fewer incidents related to AFS and massive job start
SL 5.6 test node available: tcx168;
testing Job Submission Verifier
- next iteration: transform all SGE-IO to local path(right now only default paths are transformed)

id	who	description	status/comments
old-1	NAF/ATLAS	CMT problem	request to install CERNVM-FS closed
1005-3	NAF	email notification for /scratch monitoring (instance full)	done, ATLAS will check if it is working for them
1006-6	NAF	dCache upgrade time line (10 GE in HH)	done closed
1007-1	NAF	setup Lustre groups for ATLAS, CMS, ILC	new Lustre client tested, wait for next downtime
1007-2	NAF	report on help desk tickets	shown
1010-2	NUC	resource requests	estimates shown by ATLAS
1010-4	NAF	statistics on multicore/PROOF usage	see status talk
1012-1	NAF	evaluate different stdout/stderr handling for SGE	testing JSV
1012-2	NAF	report on CERN strategy for batch and AFS	wait for Andreas H.
1012-3	NAF	new fair-share after hardware upgrade	shown in status report closed
1012-4	ILC&NAF	develop strategy to manage Lustre space	no news
1101-1	NAF/experiments	WGS reliability/support/monitoring	installed HW scanner, ATLAS gets informed on reboots
1101-2	AH/WE/HS	naf_token for ubuntu&MacOSX	no news
1101-3	NAF	revisit: user information/twitter	not working anymore, twitter changed its authentification methods
1101-4	NAF/CMS/ATLAS	move support list to request tracker?	notify YK via mail
1011-5	ATLAS	written request for NX	will be done soon
1101-6	NAF/experiments	do we need a faster x-connection?	no news
1101-7	NAF/LHCb	limit number of parallelly running jobs	some final communication needed with Andreas H.
1101-8	experiments	test new job submission verifier	tested, one issue with timeout on JSV script observed

no feedback on tickets:
- AFS Problems: rt #371712 (asked about major AFS problems on 29./30. Januar)
- motd: rt #370899

tested JSV to some degree and no problems seen so far
NX/VNC: about to write request to tech coordinators and project leader
CMT problem
- RPMs are not easy to install and mainly available for P1 releases
- cvmfs contains all releases, is easy to setup and gives good performance
- we request to install cvmfs on all NAF work group server and batch nodes

(see slides)

Wolfgang E.: InfiniBand in HH: new HW uses 10 GE? Yves K.: not yet, older machine will use InfiniBand again (maybe via IP over InfiniBand)
Wolfgang E.: Lustre performance as expected? Yves K.: performance depends on access profile
change in January minutes please have a look

1102-1	experiments	test SL5.6
1102-2	ATLAS/NAF	try CVMFS for ATLAS software

- Wednesday, March 9th 2011, 1 pm