• Although not essential to know, you might like to know that running jobs take up a special cpsbatch home area for that farm system. In fact each farm has 3 such cpsbatch home areas. The on-deck job also reserves its home area. This means that all the log files, system files, etc. is kept for the last 3 jobs run or on-deck. Hence to enable easy debugging if needed, it is recommended that we have only 3 jobs running/on-deck/queued on any one system at a time. In theory you could have tons of jobs on queue for any one system.

  • Although not essential to know, you might like to know that after a job is run, all the logfiles, oddpack files, etc. are shipped over to ea831 and is processed by factoids. Factoids is a system on ea831 that allows scripts to be run on the job files to check status, etc. and create the WEB pages on the PASS1 page.

  • Although not essential to know, you might like to know that after a job is submitted the MECCA access page is updated to show that a job is submitted. In order to make sure you are up-to-date with job submissions and completion you should always reload the MECCA access page before looking at it.
  • Log onto the host to view the job queues. If there are 2 or less jobs in a farm system you should submit more to make it up to 3 or 4. (4 if you want to leave it overnight). To view job queues do the following:
    1. Log onto a host, either fnsfu or fnckm to view systems a-e. To view system f, log onto fnio. Login as user e831p1.
    2. to view system a, use command "viewa", to view system b use command "viewb", etc. This shows up to three jobs in either the queue, on-deck, running or run-down phases.
    3. e.g. viewa (to exit use control c)

    Check Running Job Status and Farm Status

    You can check the status of a job or the farm system by doing the following:

    1. You can check the status of a job using the "showx #" command where x is the farm system, a-f,t and # is the job number for that system. You get the job number using the viewx command already described above.
      • Log onto a host, either fnsfu or fnckm to view systems a-e,t. To view system f, log onto fnio. Login as user e831p1.
      • e.g. showa 25
      • Notice the start time, time tape was mounted and whether the job should have finished since its possible for the job/system to be "stuck" sometimes.

    2. You can check the status of a farm system, i.e. to see that it is actually working and not sitting idle by using the "top#" command, where "#" is the farm system, a-f. You can also use the X-windows version of this command which is "xtop#".
      • Log onto a host, either fnsfu or fnckm to view systems a-e. To view system f, log onto fnio. Login as user e831p1.
      • e.g. topa (to exit use control c)
      • e.g. xtopa
      • Note that topf looks different as the "farm nodes" are just different CPUs on this machine.

    Staging

    Currently, all the production systems except E stage both input and output. If you need to submit a job without staging, you must type "p1submit letter tape# 0 (zero)" The only time this is normally done is when a production system runs empty. If no job is in analysis on a system, we submit the first job without staging to begin processing immediately.

    Note that system "E" knows it is not supposed to stage, so the format of the p1submit command has no influence.