Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.misc -> Re: kill sessions...

Re: kill sessions...

From: Ed Prochak <ed.prochak_at_magicinterface.com>
Date: Fri, 03 Jan 2003 04:01:04 GMT
Message-ID: <3E150E2A.80094510@magicinterface.com>


Rauf Sarwar wrote:
>
> > but why use a cannon to kill a black bird?
>
> To make sure it is pulverized on the first attempt!!!
>
> >
> > I often strongly suggest using s couple other different signals before
> > resorting to the "kill with no remorse" -9 signal. I suggest using (in
> > basically this order):
> >
> > -1
> > -2
> > -3
> > -15
> > -9
> >
> > Going straight to -9 is a bad habit to get into.
> > (I think this is the first time I ever disagreed with something Daniel
> > said.)
> >
> > Ed
>
> If the goal is to kill the session immediately....which usually is the
> case in most instances when a session is hung...then -9 signal may be
> used on the first attempt.
>
> Using other signals maybe a little mercifull but the process may
> ignore the signal entirely.
>
> Regards
> /Rauf Sarwar

Because the process may be using resources other than ORACLE. The DBA may prefer the full kill, but the UNIX Admin would prefer you try the other signals first.

Yes the process can set those signals to be ignored, but well behaved apps (at least the ones I write 8^) will process the signal and exit cleanly. Most but not all UNIX resources are cleaned up by the OS after a kill -9.

To make my example clearer, when you use a cannon to kill a bird, be prepared to deal with the colateral damage.

--
Ed Prochak
Magic Interface, Ltd.
440-498-3700(office)
Computer consulting, database and web services.
Received on Thu Jan 02 2003 - 22:01:04 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US