Re: UNIX child processes

From: <pminnis_at_sigeco.com>
Date: Wed, 02 Feb 2000 05:26:32 GMT
Message-ID: <878f65$o62$1@nnrp1.deja.com>

Our SQLNET.EXPIRE_TIME is already 10. Very nice try -- when you mentioned sniped connections, I began wondering whether a sniped connection can still leave an anonymous PL/SQL block running? This running process keeps out automated nightly shutdown from working unless we do a shutdown abort. Can anyone shed some light? I will check whether these processes are actually sniped.

In article <877r6i$a75$1_at_nnrp1.deja.com>,   ramesh_joshi_at_my-deja.com wrote:
> In article <877och$80m$1_at_nnrp1.deja.com>,
> Ben Ryan <benryan_at_my-deja.com> wrote:
> > In article <877cgr$ugv$1_at_nnrp1.deja.com>,
> > pminnis_at_sigeco.com wrote:
> > > We are running Oracle Applications on a Digital UNIX platform --
> but I
> > > have duplicated the effect with simple shell script with sqlplus
as
> > > follows:
> > >
> > > Shell script written to call sqlplus with PL/SQL block as in:
> > > when oserror exit failure rollback;
> > > when sqlerror exit failure rollback;
> > > sqlplus user/pass <<EOF
> > > define
> > > ...
> > > begin
> > > ...
> > > end;
> > > /
> > >
> > > I find the unix pid that corresponds with the sqlplus and I find
the
> > > shell script
> > >
> > > I close my telnet session
> > >
> > > Shell script process is gone. sqlplus pid continues running; it
is
> > > definitely running -- I can watch the CPU% changing up and down.
> > >
> > > This seems to be normal behavior on our system. Children continue
> to
> > > run without a parent.
> > >
> > > So my question is, why doesn't the child process go away? Is
there
> a
> > > way to subvert this behavior? Would I want to?
> > >
> > > Oracle Applications DBA's:
> > > This is occurring when we cancel a custom concurrent request.
These
> > > programs are written as host shell scripts which run sqlplus,
> perhaps
> > > multiple times in the same shell script.
> >
> > Recommend you post a minimal example shell script which will cause
> > the problem you are referring to.
> >
> > Sent via Deja.com http://www.deja.com/
> > Before you buy.
> >

>

> You have to set the following parameter in sqlnet.ora file and reload
> the listener
> SQLNET.EXPIRE_TIME= 10

>

> This feature should be supported by OS too and this can be control by
> sqlnet parameter SQLNET.EXPIRE_TIME= -- dead connection detection time
> parameter. Set a value to alert the server to ping the client to
detect
> whether the client has terminated unexpectedly and release the
> resources associated with it. The default value for sqlnet.expire_time
> is OFF. If no other value is assigned to sqlnet.expire_time in the
> active sqlnet.ora file, then dead connection detection is inactive. If
> set this parameter will release the locks but the 'sniped' connection
> has to remove manually from oracle using 'alter system kill ...' stmt.

>

> I hope this will help
> Ramesh Joshi
> joshir_at_hsn.net

>

> Sent via Deja.com http://www.deja.com/
> Before you buy.

>

Sent via Deja.com http://www.deja.com/
Before you buy. Received on Tue Feb 01 2000 - 23:26:32 CST