Title: HPSS User Forum Site Report
1HPSS User Forum - Site Report
- Bill Weeks
- Stanford Linear Accelerator Center
- Produced under contract DE-AC03-76SF00515 between
Stanford University and the Department of Energy
2BaBar The B-Factory
- Use big-bang energies to create B meson particles
- Look at collision decay products
- Answer the question where did all the
anti-matter go? - 600 physicists collaborating from gt75 sites in 10
countries - USA, Canada, China, France, Germany, Italy,
Norway, Russia, UK, Taiwan - The experiment produces large quantities of data
- 200 - 400 TBs/year for 10 years
- Heavy computational load
- 262,000 SpecInt95s available in compute farm
- 2539 compute nodes with 4181 cpus
3BaBar HPSS Software Levels
Date Machine OS DCE Encina HPSS Jul 98
F40 4.1.5 2.1 2.1 3.2 Sep 99
F50 4.2.1 2.1 2.5 4.1 Jan 01
F50 4.3.3 2.2.0.8 4.2.0.14 4.1.1.4 Aug 03
480R 8 3.2.0.4 5.0
4.5.0.1 Jan 05 480R 8
3.2.0.4 5.0 4.5.1.0
- Only change was applying 4.5.1.0 patch
- BaBar operation makes major software upgrades
difficult
4BaBar HPSS Bytes Processed per Day
5BaBar HPSS Total Space Used
6Public HPSS Software Levels
Date Machine OS DCE Encina HPSS May 01
420R 7 3.1 4.3.0.3 4.2 Dec 01 420R
8 3.1 4.3.0.4
4.3 ?????? 420R 8 3.2.0.4 5.0
4.5.1.0
- No changes for 3.5 years
- Modest activity level
7Public HPSS Bytes Processed per Day
8Public HPSS Total Space Used
9What does the future hold?
- Going it alone with Solaris
- How do we get to HPSS 6.2 system?
- What conversion steps are supported and/or
tested? - Solaris movers?
- Use mixed platforms, e.g. Linux core server,
Solaris DB2 - What versions of Linux are supported on what
platforms?
10HPSS Problems
- PVL crash during overwrite label import
- Fixed in 4.5.1.0 patch
- Tape mounting strategy when system over-committed
- Problems with drive selection, unnecessary
passthru operations - Tapes on deferred dismount queue not dismounted
first
11Summary
- HPSS is reliable
- HPSS tape handling could be improved
- Software upgrades will always be a problem