Center for Computational Molecular Science and Technology Georgia Institute of Technology Center for Computational Molecular Science and Technology School of Chemistry and Biochemistry

CCMST Weekly News, September 17 2010

September 17, 2010 1:38 pm EDT

1. Announcements
2. Statistics
3. Tip of the Week

ANNOUNCEMENTS

Physical Chemistry Seminar Series

September 21, 2010 4:00 PM – 5:00 PM MS&E 3201A
Prof. Ralph Jimenez, University of Colorado at Boulder
Ultrafast spectroscopy and microfluidics-based methods for investigating protein dynamics

STATISTICS

FGATE

Uptime: 37 days
/home directory usage: 73% (1.6 TB available)
/backups directory usage: 84%

LSF usage for Week 36 (9/6-9/12) (times are in minutes)
GroupJobsTotal CPUAvg CPU Avg WaitAvg Trnr.
Bredas 59 117612 6% 1993 595 4190
Hernandez 1875 477036 25% 254 50 305
Sherrill 183 378039 20% 2066 10502 12573
Other 3 25307 1% 8436 144 8011
Total 2120 997989 52% 471 967 1483

Note: percentages refer to the total CPU time available for the period.

Most productive user of the Week: kennedy 378005.

EGATE

Uptime: 292 days
/theoryfs/common directory usage: 36% (428GB available)
/theoryfs/ccmst directory usage: 85% (136 GB available)

LSF usage for Week 36 (9/6-9/12) (times are in minutes)
GroupJobsTotal CPUAvg CPUAvg WaitAvg Trnr.
Hernandez 225 60190 4% 268 1 276
Sherrill 960 417861 28% 435 42 436
Other 26 34967 2% 1345 0 1351
Total 1211 513018 34% 424 34 426

Note: percentages refer to the total CPU time available for the period.

Most productive user of the Week: sahan 233638.

TIP OF THE WEEK

By Massimo

Cluster-fork

Cluster-fork is an utility that can be used to execute commands on the compute nodes of a cluster managed by Rocks (i.e. egate and fgate). For example, to list all your processes  on the compute nodes of the cluster type:

  • cluster-fork ps -U$USER

Cluster-fork uses ssh to launch the chosen task serially on each compute node, ignoring the nodes that are down. By default the job is blocking: cluster-fork will wait for the command on one node to be executed before starting the job on another node. This behavior can be altered with the --bg option. This will start the jobs in the background:

  • cluster-fork --bg hostname

To select the cluster nodes on which to execute the command, use the --nodes option followed by the list of compute nodes where to run the command. A range of nodes can be specified with a shorhand notation:

  • Continuous range: --nodes=compute-0-%d:0-2 -->  compute-0-0 compute-0-1 compute-0-2
  • Discontinuous range: compute-0-%d0,2-3 --> compute-0-0 compute-0-2 compute-0-3
  • Multiple elements: compute-0-%d:0-1 compute-1-%d:0-1 --> compute-0-0 compute-0-1 compute-1-0 compute-1-1

Do you have usage tips that you want to share with the other CCMST users? Please send them to Massimo (massimo.malagoli@chemistry.gatech.edu) for inclusion in the Tip of the Week section.