How about if we use Dead Connection Detection suing SQL_TIMEOUT
parameter in sqlnet.ora file.
On 6/9/05, Paul Drake <bdbafh_at_gmail.com> wrote:
> On 6/9/05, Greg Norris <spikey.mcmarbles_at_gmail.com> wrote:
> > Perhaps assigning an idle-timeout in the user profile(s) would be easie=
r?
> >=3D20
> > On 6/9/05, Onkar N Tiwary <onkarnath.tiwary_at_gmail.com> wrote:
> > > hi oracle gurus,
> > > I am running oracle 8i(8.1.7.4.0.0) on windows 2000 adv.server ( 2 s=
er=3D
> ve=3D3D
> > r
> > > is in remote area). Everyday in the morning I used to find hell a lot=
o=3D
> f
> > > INACTIVE users of last day. So I had to kill their sessions and then =
ki=3D
> ll
> > > their threads using ORAKILL. My question is:
> > > 1) Do anyone has got the script to kill the threads automatically?=
=3D3D2=3D
> 0
>=20
> Greg,
>=20
> (speaking in terms of Oracle 8i R3 and 9i R2 on win32)
>=20
> Idle-timeout in the user's profile only "SNIPE"s a session, it doesn't ki=
ll=3D
> it.
> The user is notified that the session was killed after they attempt to
> make another call. If the user does not make another call in that
> session, the session does not get killed, hence another method is
> required to perform garbage collection.
>=20
> This is even more fun with regard to shared server configurations.
>=20
> In a related matter, dead_connection_detection was supposed to work in
> the 9.2.0.5 patchset, but I don't believe that it did.
>=20
> Fundamentally, having user shadow processes as threads under the
> oracle.exe process prevents the use of timers as there is a limit of
> 16 timers per process.
>=20
> I haven't checked to see if the functionality in 10g R1 is there yet
> but will do so this week.
>=20
> Paul
> --
> http://www.freelists.org/webpage/oracle-l
>
--
http://www.freelists.org/webpage/oracle-l
Received on Wed Jun 15 2005 - 12:07:00 CDT
