Re: Dispatcher freeze

This is a known bug, fixed in 8.0.5.1 (or 8.0.5.2?) It bit us for the longest time, and it took several tech support people before we got someone who recognized the bug and told us to upgrade.

A few stopgap solutions, you can try turning off Connection Pooling if it's on. And you can try increasing the number and size of redo logs (one tech support person seemed to think this bug was related to an separate problem, i'm dubious).

Also here's a quick way to check all the dispatchers: for i in $(lsnrctl services | grep PORT=) ; do  tnsping "$i" | grep OK >/dev/null || echo "$i is dead" done

You can kill -9 a dispatcher and Oracle should deal properly.

"quidam17" <quidam17_at_nospam.hotmail.com> writes:

> First of all, I apologize for my english. This is not my native language...
>
> We have a HP-UX 10.20 server with Oracle InterOffice on a Enterprise Edition
> Server 8.0.4. Of course, we use MTS option with 20 dispatchers since we have
> over 600 mail clients logged on each day.
>
> On a regular basis (2-3 times per week), a dispatcher become freeze... Since
> this is probably the least used dispatcher, Oracle try to use it again when
> a new attemp to connect is made and then, the new connection fail.
>
> If I kill the listener, all new connections will be in dedicated mode...
> If I can identified the frozen dispatcher and killed it, Oracle restart it
> and we are back in business with MTS.
>
> It seems that the OS on HP server is up to date with all patches.
>
> Anyone have idea on what cause this problem...?
>
> Thanx!
>
> Mario Perreault
> DBA, Ville de Québec
 

-- 
greg