yong321_at_yahoo.com wrote:
> On Jan 31, 6:48 am, GS <g..._at_canada.com> wrote:
>
>>The server is running Windows 2003 advanced server, 8GB RAM, 8 3Ghz Xeon
>>processors, /fastdetect and /PAE switches are in boot.ini
>>
>>Database is 9.2.0.6, shared pool 400 MB, buffer cache 800MB, large pool
>>8MB SGA max size 1505 MB - Typical number of connected users is between
>>75 and 125, running in dedicated server mode. Aggregate PGA target is
>>150MB, and application connects to database via ODBC. Processes is set
>>at 150, I was thinking of increasing this but when I looked at the max
>>utilization when these errors were happening it was only 136.
>>
>>The database has been running since Sept with no problems, until a few
>>days ago when people suddenly could not connect. Looking at listener.log
>>showed a lot of TNS-12540: TNS:internal limit restriction exceeded and
>>TNS-00510: Internal limit restriction exceeded messages, along with some
>>TNS-12500 and 12560's. The number of users connected was between 120 and
>>130, and nothing out of the ordinary was being done. I bounced the DB to
>>clear the problem and it has not happened since, and no connection
>>errors are showing in listener.log
>>
>>Since many connections are idle for long periods of time, I was
>>considering implementing MTS (or maybe DCD), but I thought that one
>>shouldn't need this (MTS) until connected users hit closer to the 1000 mark.
>>
>>Has anyone here switched to MTS, and if so, what threshold did you use
>>to make this decision?
>>
>>Or should I be looking elsewhere to find the problem and/or cure?
>>
>>thanks!
>
>
> You hit ORA-12500 maybe too early considering you have 8GB RAM. What's
> the complete error? I'm looking for something like this:
>
> TNS-12500: TNS:listener failed to start a dedicated server process
> TNS-12540: TNS:internal limit restriction exceeded
> TNS-12560: TNS:protocol adapter error
> TNS-00510: Internal limit restriction exceeded
> 32-bit Windows Error: 8: Exec format error
>
> Next time when the error occurs, quickly check Virtual Bytes in
> perfmon for oracle.exe process before you bounce.
>
> Yong Huang
>
I thought it hit that too early as well, and I thought that the /PAE
switch should have allowed Oracle to use more RAM if neccesary as well,
but perhaps I am mistaken.
The errors in the listener.log are exactly as you posted. I will set up
a perfmon window and watch the oracle virtual bytes and see what happens.
Received on Wed Jan 31 2007 - 09:05:24 CST
