Re: orphan shadow tasks

From: Joe Minnich <jminnich_at_bugaboo.ssnet.com>
Date: 10 Oct 1993 05:37:43 GMT
Message-ID: <298737$7t1_at_marlin.ssnet.com>


>Has anyone run into this problem and licked it ? Would be real
>interested in hearing about others experience with this.
 

>> We just had oracle support onsite for 3 days. Apparently, if someone
>> "Control-/"'s out of a form, it leaves a "shadow" process. I'd guess a
>> PC rebooting would cause the same effect.
 

>>(This is on a Sequent Symmetry, Oracle 7, Forms 3.0)
 

>>Recommendation was: "Don't 'Control-/'". (*sigh*) - tell that to your
>>users. Our workaround is a cron script that goes on a search and destroy
>>mission for all "oracle<database>" processes with a ppid of 1.
>>--
>>Jonathan

Thanks for the feedback John. This is a real hoser from Oracle. It's easier to workaround if your originating ppid is a user unix process. init grabs the orphan shadow and at least you can identify and kill them in an automated fashion. I am not sure this is the case in a client server environment. I think the shadow task retains its originating ppid when it is forked via the dos client connection request. The parent process is probably the orasrv or starsrv listener process. Hence i don't think the shadow tasks are being adopted by init. If we can find a way to indentify the orphans we can probably use a similar approach.

thanks again for you help.


joseph r minnich	jminnich_at_bugaboo.ssnet.com
724 millcreek lane	home: (302)322-6244
bear delaware 19701	work: (302)368-8982
    
Received on Sun Oct 10 1993 - 06:37:43 CET

Original text of this message