Well, have you tried it?
In my understanding orakill is actually an OS thread
killer written by Oracle, it just searches the appropriate oracle.exe based
on SID and kills the thread with OS thread id in it. It does
not log on to Oracle to check whether Oracle knows anything about this
thread or not, it just simply kills the thread with id you provide and that's
it.
Tanel.
----- Original Message -----
Sent: Thursday, September 18, 2003 4:05
PM
Subject: Re: spurious threads
We know the thread id from running pslist from
sysinternals.com. Orakill will not kill it probably since doesn't think
it really exists.
>>> [EMAIL PROTECTED] 9/18/03
9:49:41 AM >>>
Hi!
Orakill is able to kill threads based on thread id. But
since you can't find out the "runaway" thread id-s from v$process, you could
find all thread id-s in oracle.exe, minus those spids which exist in
v$process and kill those. But I definitely recommend to turn to support first,
because killing threads that way isn't a nice thing to do in production system
and might be good enough only for temporary workaround, not for continous
usage.
Tanel.
----- Original Message -----
Sent: Thursday, September 18, 2003 4:19
PM
Subject: spurious threads
We have had an ongoing problem for nearly 1.5 years now starting with
8.1.7.3.2. If we patch the Oracle Apps 11i, a majority of the time,
the oracle.exe process gets a runaway thread. A thread exists in the
oracle.exe but not in v$process/v$session. Shutting down the database
does not resolve the problem. Only stopping the service will resolve
it.
Two weeks ago implemented FND G/HRMS F in dev/test with production
scheduled for this weekend. Since then have noticed high thread
counts. Dev has topped out at 1000+, test is currently at 809 with
only 50 v$process entries!!!!
Help.
Jeffrey Beckstrom Database Administrator Greater Cleveland
Regional Transit Authority 1240 W. 6th Street Cleveland, Ohio
44113 (216) 781-4204
|