Re: Problems with MTS.

"John Peterson" <johnp_at_azstarnet.com> wrote in message news:teu5i66l87e68d_at_corp.supernews.com...
> Hello, all!
>
> We are running Oracle 8.1.7 on RH Linux 6.2 (I think ;-). Recently, we
> tried activating the Multi-Threaded Server (MTS) capability of Oracle to
 see
> if we might get a better connection/process pooling capability.
>
> In our sandbox, when trying to execute a regular query with SQL*Plus (from
> an 8.1.5 installation, though) it just seemed to "hang". We could
 seemingly
> connect just fine, but the query just never returned.
>
> When we turn of the MTS feature, then SQL*Plus can run the query
> successfully.
>
> I understand that this is a broad question, but does anyone know where we
> might concentrate our troubleshooting efforts with MTS?
>
> Any ideas/suggestions would be much appreciated! Thank you! :-)
>
> John Peterson
>
>

Hi John,

I have been troubleshooting this same problem at a big banking firm in the Netherlands as consultant, so I studied all aspects of MTS and I was unable to reproduce the problem, once *I* had set MTS up. Looks like the ratio between dispatchers and servers has been wrong. Can you post and/or e-mail your mts init.ora parameters and output from v$dispatcher, v$mts, v$circuit and v$shared_servers. I might run into something similar in the near future, and I am *anxious* to know what is going on.
In 8.1.6.1 for Solaris there are situations where the dispatcher hangs for *exactly* 60 seconds, this has been qualified as a bug by Oracle and is resolved in 8.1.7 for Solaris. I'm not sure whether this is a generic bug, it hasn't been qualified as such by Oracle. I would go for an adequate number of servers first. This means for me : 2 dispatchers, and at least 10 servers

Also : did you ever monitored your session using v$session_events and/or similar views? What was the outcome. It also looks like an incorrect relationship between the session data unit of net8 and the MTU of the server will be killing things. It also looks like not performing array fetches is killing performance

Hth,

Sybrand Bakker, Oracle DBA Received on Tue May 01 2001 - 16:02:25 CDT