Using the CISER Research Nodes:  Managing Programs and Processes on the CISER Research Cluster

ABOUT RUNNING PROGRAMS:

• The applications (SAS, SPSS, Stata, etc) on any node are independent of those same applications on any other node.  

  • Therefore output created on one node must be saved to your home directory in order to be accessible by any other node.
• You must be logged on to an CISER Research node in order to run applications there.

  • If you attempt to run a program file on the CISER file server from a drive mapped to your local PC --- it will search for and run the relevant application on your local PC

DIRECTING PROGRAM OUTPUT:

• Be sure to direct output from your programs to your home directory on the CISER file server ("ciserdfs1" mapped as U:\UserX\your netid) or to temporary scratch space  (the "T:" drive on any node).  Do not attempt to write to the "C:" drive of the CISER Research nodes.

  • See How to Manage Your Files for more detailed information about disk space.

MANAGING PROCESSES ON THE NODE TO WHICH YOU ARE LOGGED ON:

• Use Task Manager on any node to identify and manage the applications you have running on that node.
  • Start --> Run --> open "taskmgr" --> OK   (or right-click on an empty space on the taskbar and choose "Task Manager")
    
  • Choose the "Applications" tab and you will see your current tasks and their status (i.e. "running", "not responding", etc)

    • You can end tasks that are "not responding" by highlighting the relevant task and choosing "End Task".
  • Options available under the "Processes" tab will allow you to see a list of your current processes, monitor CPU usage, and end processes if you choose.
  • By choosing the "Performance" tab in the Task Manager window you can see how much total memory and CPU are being used on that node (by all users). 

MANAGING PROCESSES ON MULTIPLE NODES IN THE CISER RESEARCH DOMAIN:

• Use Terminal Services Manager on any CISER Research node to identify and manage the applications you have running on all nodes in the CISER Research Domain:

  • NOTE:  This is an excellent way to locate and log off any disconnected sessions.
    • See Disconnecting vs Logging Off for information about why this is important!
  • INSTRUCTIONS:
    
    • From the current node:
      • Start --> All Programs --> "Terminal Services Manager"
    • From the Terminal Services Manager window:
      • Click on the "+" in front of "All Listed Servers".
      • Double-click on "CISERRSCH".
      • Highlight the node of interest on the left hand side of the window and choose the "users" tab on the right.  You will see a list of users with current active and disconnected sessions.
      • If you see disconnected sessions listed under your own userid, and you do not have jobs running there,  you can right-click on that session and choose "log off".   

• MANAGING MULTIPLE SESSIONS ON ONE OR MORE NODES:
  • You may have multiple sessions open on the same or different CISER Research nodes at any time.  These sessions may be either "active" or "disconnected".   Disconnected sessions are those for which you have closed your Terminal Services Window but have not logged off.  You should log completely off when you close your Terminal Services Window unless you have have jobs running.  It is up to the user to keep up with these sessions and completely log off sessions not in use.

  • To check the status of your current sessions on any CISER Research node:
    • Go to "Start" -->"Programs" --> "Administrative Tools" --> "Terminal Services Manager".
    • Options available in the "Terminal Services Manager" window will allow you to list and manage your processes and sessions on all of the CISER Research nodes. You can log off from disconnected sessions on other nodes by selecting the node in the left-hand column, and then right-clicking on the session id with your netid on the right-hand column and choosing "log off".

NEED ACCESS TO MORE RESOURCES?

If you need access to more resources than the CISER Research nodes currently have to offer (to do parallel processing, for example, or run Fortran or C++ code, etc) you may want to access to the  Cornell Center for Advanced Computing Resources.