use Perl

• You may need to do more work than that (Score:1)

  by btilly (5037)

  One of the signals that can kill the Java program is a kill -9. Which can't be caught. Ever.
  
  If your sysadmins are being happy with kill -9, then what you may need to do is write a monitoring process which is told what each Java process is doing, and will tell Oracle to clean up after them when they disappear without properly cleaning up.
  
  Another possibility is that the bug is in Oracle. Unlikely I know, but back in the Oracle 8 series I ran across a query that caused an internal Oracle process to try to f
  • Re: (Score:2)

    by jdavidb (1361)

    Yeah, that's why I said "catch all signals possible." :)

    Part of the issue is actually that I myself and one of our monitoring groups have to kill this every so often, although I definitely don't kill with 9 unless something refuses to go away with something weaker.

    I'm not sure a second process could tell Oracle to release these locks, because it wouldn't own them. So I'm not sure the monitoring process idea will work. It might be possible, however, for an audit process to run on the server side and ki

    --
    J. David works really hard, has a passion for writing good software, and knows many of the world's best Perl programmers
    • Re:You may need to do more work than that (Score:1)

      by btilly (5037) on 2007.07.27 10:14 (#56549) Journal

      You'll have to work with a DBA on how to do it, but it is possible for a second process with sufficient privileges to kill someone else's login on the Oracle side. When that login is killed, all locks that it held are released.
      
      This does take a lot of work to set up though. I wouldn't go there unless you really have to. But be aware that it is an option.

      Reply to This

      Parent