Birmingham
From GridPPwiki
Birmingham Particle Physics Group (http://www.ep.ph.bham.ac.uk/)
| Table of contents |
[edit]
Our Grid Resources
[edit]
Atlas cluster
- Number of Job Slots: 28
- CPU Model: Intel xeon
- CPU speed: 2GHz
- SpecInt: 756
- SpecFloat: 764
- RAM: 1024MB
- Supported VOs: alice atlas babar biomed calice cms dteam hone ilc lhcb sixt zeus
[edit]
BaBar cluster
The Grid part of this farm is down due to ongoing problems with power supplies and several broekn disks.
- 1 one node for a BaBar xrootd server
- 1 one node for a BaBar Objectivity server
- 5 nodes dedicated to the PPS
- Number of Job Slots: 104 (4 job slots lost because of faulty disks)
- CPU Model: Intel PIII
- CPU speed: 800MHz
- SpecInt: 340
- SpecFloat: 226
- RAM: 1024MB
- Supported VOs: alice babar cms hone lhcb zeus (90% of jobs slots for BaBar)
[edit]
Storage
- 1.8TB raid array lying behind DPM/SRM
[edit]
Monitoring and Accounting
- Ganglia Status (http://epgmo1.ph.bham.ac.uk:8888/ganglia)
- gstat Monitoring (http://goc.grid.sinica.edu.tw/gstat/BHAM-LCG2/)
- Apel Accounting (http://goc.grid-support.ac.uk/gridsite/accounting/tree/gridpp_view.php?ExecutingSite=BHAM-LCG2)
[edit]
Networking
Answers to GridPP 'Ten Easy Network Questions': Birmingham (10 Questions)
[edit]
Addition of extra clusters to an existing LCG site
There are three ways to add an extra cluster to an existing LCG site, namely:
- Using CPU scaling
- easy to implement, but
- only the CPUs characteristics of the clusters can differ
- the same VOs must be supported on all clusters
- the site information provider does not reflect the true physical caracteristics of the clusters
- Adding a new CE
- the most versatile solution
- not neat for a site with many small clusters (for example with 4 clusters with 10 nodes each)
- implemented at Bham, see Addition of a second CE at the BHAM-LCG2 site for instructions
- Hosting several clusters under one CE making use the of GLUE sub-cluster attribute
- Ideal for small site with many farms
- allows to enable different VOs on different clusters
- neat use of the GLUE schema, but
- APEL requires to add a prefix to the site name
- got it working, but chose solution 2. for the time being
- I will put some instructions in the wiki for information in the not too distant future
[edit]
Our experiences
[edit]
Migration from a classic to a DPM/SRM SE
[edit]
Enabling the DNS VO style test.gridpp.ac.uk
[edit]
gLite Update 27
[edit]
Ongoing Work and Plans
- Upgrade to LCG-2_7_0
- Fix and add another 40 nodes to the BaBar farm
- Integration work of recently deployed LCG-2_7_0 WNs, MON and RB in the Pre-Production Service
- Birmingham in Service Challenge 4
