Tier-2 Weekly Report 2009-02-10

SAM Results

• Source URL: http://lcg-sam.cern.ch:8080/reports/t2/site_avail.xsql

• ops-availability-20090210.png:
  

• ops-reliability-20090210.png:
  

• atlas-availability-20090210.png:
  

• atlas-reliability-20090210.png:
  

APEL Accounting Publishing Status

• apel-gant-20090210.png:
  

APEL Accounting Results

• apel-table-20090210.png:
  

General News

• 20090202 - backlog of panda jobs at T2s that are in a state of 'transfering' of their output to TRIUMF Tier1. Resolved 20090205 with new install of site services (at CERN) and cloud reactivated
  • ADC Ops Elog https://prod-grid-logger.cern.ch/elog/ATLAS+Computer+Operations+Logbook/2074
• next round of distributed analysis challenge for CA cloud scheduled to commence 2009-02-10 at 09:00 PST (baseline and FileStager tests)
  • http://gangarobot.cern.ch/st/test_132/
  • additional datasets subscribed starting 2009-02-06 to Victoria and Toronto to bring number of input datasets (jobs) closer to 200-level
• draft January 2009 accounting statistics published; CA-East 87524 kSI2k-Hrs (98% of pledge); CA-West 173992 kSI2k-Hrs (130% of pledge)

Site Reports

Alberta

Outstanding and Recent Issues

• no real issues
• few WNs to be debugged with respect to missing stderr/stdout
• still need to add 2nd GigE nics to dCache pools for internal network access from WNs to avoid NAT
• installed and tested siteTests package [2009-02-07]

• testing HEPSPEC version of SpecCPU2006 on nodes for WestGrid-T2 preparations [2009-02-02]
  • preliminary comparison of default SL47 (2.6.9-78.0.13.ELsmp) and up-to-date kernels (2.6.28.2) with 32-bit code test (64-bit OS)
    • Machine = Dell PE1950 2 x 4-core Intel Xeon E5405 @ 2.00GHz, 16 GB memory
      • kernel = 2.6.9-78.0.13.ELsmp: HEP-SPEC2006 score = 55.65
      • kernel = 2.6.28.2 : HEP-SPEC2006 score = 55.86

SFU

Outstanding and Recent Issues

• Ganga jobs failing at SFU a large fraction of the time and outstanding since 2008-11-10 - under ongoing investigation (will likely see effects in DA stress tests as well)

• SFU analysis and production pilot logs empty starting 2009-02-05 coinciding with change to ldap authentication but was due to failing dcache door ; passwd auth restored temporarily
• hostcert expired for snowpatch [MON services host + torque/moab] [2009-02-05]

Victoria

Outstanding and Recent Issues

• firewall issue was blocking site apel statistics reporting since around mid-January; resolved [2009-02-03]
• A. Agarwal now away on holiday for several weeks; local admins keeping watch (and A. Agarwal remotely when possible)

Toronto

Outstanding and Recent Issues

• increased size of MCDISK to 9.09 TB since it was close to being full [2009-02-09]
  • Occupied TB: 7.00741812728 Total TB:7.27595761418

TRIUMF

Outstanding and Recent Issues

• CA/TRIUMF pilot put error copying output from WN to localSE - resolved [2009-02-10]
  • https://prod-grid-logger.cern.ch/elog/ATLAS+Computer+Operations+Logbook/2136
• FTS server upgrade [2009-02-05]
• new round of pre-stage tests started [2009-02-09]

Production

Summary

site	success	failure	success (walltime)	failure (walltime)	efficiency	efficiency (walltime)
TRIUMF-LCG2	21706	1135	262201110	11445124	95%	95.8%
SFU-LCG2	3112	348	48997010	3833158	89.9%	92.7%
VICTORIA-LCG2	2545	62	50463362	612398	97.6%	98.8%
ALBERTA-LCG2	1707	107	31979725	2061240	94.1%	93.9%
TORONTO-LCG2	1573	105	72600590	1182258	93.7%	98.4%
total	30643	1757	466241797	19134178	94.6%	96.1%

ProductionDashboard.pdf

Failed Jobs

ProductionFailedjobs-10Feb09.pdf

Notes

The numbers in the Summary page do not seem to match the Failed Jobs page for the Tier-2s; this could be because of the failed Transfer jobs which timed out after 2 weeks.

-- AsokaDeSilva - 10 Feb 2009