Running cluster jobs from Windows requires 3 tools (curtesy of GW).

  1. Putty - a free telnet SSH client (whats getty?)
  2. WINCSP - a free ftp and transfer client
  3. Cygwin/X - a free X-Windows client (activate XWindows if you use setup.exe from cygwin)

...don't forget to start your Cygwin bash shell and enter startx
...enter your login name and pwd and enter the clustername in Putty and WINSCP, setup X11 forwarding in PUTTY, voila!

The Sun Grid Engine can be configured via qmon
The 8 GByte nodes on Shiraz can be accessed via qmon -q highmem.q
The Explorer style File Manager can be started on the head node with nautilus
See other explanations in this PPT.

Command Description
qacct Extract accounting information from cluster
qalter Changes the attributes of submitted but pending jobs
qconf SGE's cluster, queue etc configuration
qdel Job deletion
qhold Holds back submitted jobs for execution
qhost Shows status information about SGE hosts
qmod Modify queue statues: enabled or suspended
qmon X-windows Motif interface
qrsh SGE queue based rsh facility
qselect List queue matching selection criteria
qsh Opens an interactive shell on a low-loaded hosts
qstat Status listing of jobs and queues
qsub Commandline interface to submit jobs to SGE
qtcsh SGE queue based TCSH facility
qtcsh, qsh extended command shells that can transparently distribute execution of programs/applications to least loaded hosts via SGE.

Some general command line tools for UNIX and LIN.X

find something in files find . -print | xargs grep "findmestring"
list files ls -l
which java version which java
which java compiler which javac
show free memory free
show detailed memory info cat /proc/meminfo
show number of CPUs or processors cat /proc/cpuinfo|grep processor|wc -l
show CPU type cat /proc/cpuinfo
show running programs or tasks ps aux|sort -n +2
show tasks ps aux|sort -rn +2|head -10
history show command history you typed
env show environment variables
show processor and memory load 5x vmstat 2 5
show every 2 sec. cpu san mem info vmstat 2
show running tasks top
show running tasks from user B. Gates top -u bgates
check largest memory pages: cat /proc/meminfo | grep Huge
startx start X-Windows
break some command or task Ctrl-C or Strg-C
show free space on disk or HD df

We have access to two clusters.


  • 111 nodes, dual socket dual core Opteron running Rocks 4.1
  • 2 Opteron 270 (2.0 GHz) processors per node
  • 37 nodes w/ 8GB Ram, the rest have 4GB per node
  • 6.4 TB (raw) backed up redundant storage


  • 24 nodes, dual Opteron running Rocks 4.1
  • Dual Opteron 248 (2.2 GHz) processors, 4GB RAM per node
  • 3.2 TB (raw) backed up redundant storage