Manchester HEP

1
Manchester HEP

• Desktop/ Laptop
• 30 Desktop running RH 7.3
• 20 Laptop Windows XP RH OS X
• Home server AFS using openafs 3 DB servers
• Kerberos 4 we will move to kerberos 5 by sept.
• Web server AFS
• Mail Server NFS
• Backup
• We used Kickstart for the installation
• Security
• 5 Farms
• Storage Space 12 TB

2
Software

• D0 Software mounted via NFS to the farm
• Atlas Software mounted via NFS to the farm
• CERN Software via AFS

3
Network
Default deny
Default allow
Default Deny
4
Farms
Farms

• Farm1
• consists of 40 dual-processor worker nodes
• with 800 MHz PIII processors and 1 GB of memory.
  
• Two gatekeeper machines, bfa and bfb, provide
  front-end
• access and NFS servers for the farm. The PBS
  batch system is
• used to direct jobs to the worker nodes, with a
  maximum of 2
• jobs per dual-processor worker node.
• Gatekeepers and worker nodes are installed using
  the AFS
• Cluster's kickstart procedure, and are very
  similar to the
• Redhat Linux 7.3 desktop machines. Although full
  access to
• AFS is available, temporary

5

• Farm2
• 30 dual-processor PCs
• 2GH AMD
• Used for D0 and Atlas.
• 2 GB Memory
• PBS batch system is used to direct jobs to the
• worker nodes, with a maximum of 2 jobs per
  dual-processor worker node.
• There is a SAM station attached to this farm.
• and we running JIM which is much easy and
• user friendly than LCG.

6
Farm 3 Apple G5 No need for Linux
This is new 6 X-server with 2.8TB Disks We
are using it to run simulation work have
started testing Atlas and D0 simulation and
analysis
Farm 4 16 Single processor Used for Grid Testbed
7
Farm 5 Tier 2 Farm 26 xpc Shuttle 13 P4
machines 3GHz 13 AMD 64 bit machines 3.2 GHz
Future
1500 Nodes to come with 200 TB via Dcache Will
be installed with LCG software
8
Problems 1- Air - conditions 2- Power to the
machine Room 4- We had problems with AFS