Minutes of July 28, 2003 CD Operations
- No report.
- Data Handling: Last week dCache moved an average of 20 TB/day (100 Million
Events/day) with a peak rate of 24 TB (120 Million events) on Friday.
The average rate was 10% higher than the previous week.
- CAF: over the last week the fraction of CPUs utilized has averaged to
around 75%, down from 80% last week.
- The farms are running really well. Over the last week they processed
45 million events: an average of 6.5 million events/day. They are now
within 5 million events of the recent data taking, which compares
well with the beginning of last week we when we were 30 million events
- About to deploy LCG computing site.
- Metrics were shown for farms processing, event consumption, TB consumption,
event size on analysis stations, CPU usage, CPU time/Wall time.
Farm: 14.3 M events collected, 12.27 M processessed. Running p14.03.01.
Seeing some competition for tape drives with user access of raw data. Will
add more drives next downtime--thanks for CCF for loan of mover nodes.
p14 timing comparable to p13 timing, but p14 somewhat less robust.
We have "signed off" on the dst level reprocessing at NPACI for the test
48 TB 9940b (wrote 8 TB last week--some what more than expected)
92TB in LTO
Remember--D0's primary analysis format is the thumbnail--20kbytes/event.
Projects Data Analyzed Events Analysed Data transferred in
D0mino 457 18 TB 165M 10 TB
CAB 621 22 TB 900M 10 TB
Clued0 226 2 TB 13M 1 TB
D0Karlsruhe 47 0.5 TB 18M 0.5 TB
Tape access was causing problems with intra-station cab transfers.
Changed station parameters to reduce the number of concurrent enstore
transfers to a node, and asked the users who were accessing raw data to
back off a little until after L-P.
- Paper claiming confirmation of dark energy. Press release showed
up at NPR web site, with Albert Stebbins giving an audio presentation.
Bob Nicholl at U. of Pittsburgh will appear on Monday's David Letterman.
- SDSS DB2 and 3 are gone.
- Data processing went from 79% to 77%, because now using a new release.
- Minor problem with XTape in OCS on Linux.
- No report.
- Tuesdays upgrading microcode on D0 robots from 8-12, and will commission
9 new T9440B drives.
- Wednesday downtime for CDF to upgrade microcode proposed. Want to delay
this until Lepton-Photon underway.
- Scheduled maintenance Thrusday 6-7 a.m.
Border routher module rearrangement & upgrade, switch upgrade.
- Increasing time of blockage for inbound scanning.
- Finished phase 1 of DAB fiber to desktop upgrade.
- Finished work on link loss problems on farms.
- Conference rooms: meeting with D0 about DAB conference room requirements.
- Offsite: conference call with MRB communications.
- Tomorrow blocking unregistered onsite web servers from serving pages offsite.
1. Atipa is replacing all the seagate 80 gig drives in the 240 D0 nodes, due
to many disk errors. The drives are less than a year old.
1. Ken and Andy installed the latest security patches on the
fermi production domain controllers.
2. PPDserver was down Saturday morning (scheduled). A
drive chassis was added and the BIOS and firmware
were upgraded. 3 hours.
- Recycler ring bpm project near completion.
- Tevatron bpm project ramping up.
- BTeV pixel group ready for 6 plane detectors.
Planning and Customer Support
- Tomorrow morning budget meeting 11-12.
- Goals: we have 219 out of 258.
- Budget planning for FY04 on Monday Aug.4.
- Number of open problems down.
- WH power outage on Saturday.
- Computing tecniques seminar on Tuesday Aug. 5 on computing using
- Data center cleaning in progress.
- UPS is hear for installation.
- Plans for CD tours of FCC for Lepton-Photon discussed and handed out.