HP OpenVMS Systems

tracking cause of varying elapsed time on job?

» close window

The Question is:

 
I am having trouble with a batch job that varies in the amount of time it
takes to process.  One night it will run in 2 hours, other nights it can take
16 hours!  What  is the best manner to monitor this job and determine what
resources it may need and/or what may be slowing it down?
 
Also, what is the best way to determine whether the image it is using should
be installed or not.
 
Thanks!
 
 
 
 

The Answer is :

 
  There is no best manner -- what you need to do is observe what it is
  doing, and what it is waiting for.  Assuming the batch job always runs
  from the same username and the quotas on the username and the queues
  are not being changed, the cause could be lock-related, contention on
  a disk spindles, insufficient physical memory, or it could be contention
  for locked records in a file.  There are likely other potential causes.
 
  You could use a variety of profiling tools, or you could consider the
  integration of debugging code directly into the application, code that
  allows you to see what the application is doing and when...
 
  DECamds is one option for watching lock, I/O, and memory contention.
  (DECamds is licensed with OpenVMS in recent releases and with clusters
  in older releases, and is seperately installed.)
 
  Also see the OpenVMS performance manual for related information.

  

     

     
answer written or last revised on ( 16-JUL-1998 )
     
» close window