Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: HELP: ORA-12500 error, SQL*Net V2

Re: HELP: ORA-12500 error, SQL*Net V2

From: <oliver.willandsen_at_sg.cec.be>
Date: Wed, 05 Aug 1998 13:36:57 GMT
Message-ID: <6q9n5q$oh9$1@nnrp1.dejanews.com>


In article <6q9fo7$egf$1_at_nnrp1.dejanews.com>,   tgp_at_mci2000.com wrote:
>
>
> > Well, get back to me if I'm wrong but I do think it's rather obvious
> > especially since (>It's not solid, by the way. > Sometimes you get it and
> > sometimes you don't.)
> >
> > I think it's quite simply the number UNIX processes AIX can spawn. If that
> > value is set to anything lower than the 250 you allow for Oracle users, each
> > time a new user tries to connect Oracle tries to fork off a process and
> > can't.
> >
> > Use SMIT to change the value online - no need to reboot
>
> Well, you sure make it sound obvious, but ...
>
> 1. This problem is occurring on our test machine, which has far fewer users
> and processes than our production machine, which is configured just like it.
> A "ps" command shows about 70 processes on the machine that exhibits the
> problem, and about 200 on the machine that's healthy.
>
> 2. Also, I am not aware of an AIX parameter that sets the global max process
> count. There is certainly a per-user maximum, but I have never found a global
> one. If you know how to actually get one's hands on this thing, I'd like to
> know about it.
>
> 3. The dedicated server process IS being forked. It is just getting a "signal
> 11" and dying. The listener.log shows the listener is seeing an EPIPE error,
> too. I interpret to mean that the listener is forking a dedicated process,
> (not getting an error on the fork), setting up a pipe to the process, and
> then later being notified by AIX that the other end of the pipe is dead.
>
> -----== Posted via Deja News, The Leader in Internet Discussion ==-----
> http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum
>

OK, maybe it's not quite _that_ obvious.

If you're Oracle setup is as a dedicated server - which is the case with us - and not multi-threaded ( which our Oracle Support doesn't recommend ), Oracle will try to fork a new process for each user trying to connect. This is where the maximum number of processes _per user_ comes in - so you're right - there is no global max process count - but the problem is not global. The listener is seen a _one_ user.

So check how your Oracle system is set up dedicated server or multi-threaded.

Then, if you have SMIT do smit -C and look in [System Environments], then in [Change / Show Characteristics of Operating System] - change Maximum number of PROCESSES allowed per user to a value - check if the problem persist.

We too have a number of different machines development, production etc. The PROCESSES value on our production machine is 350, but only 150 on the development machine - which makes sense.

HTH --
Oliver Willandsen - European Commission http://europa.eu.int
All remarks are my own and do not necessarily reflect official European Commission policy

-----== Posted via Deja News, The Leader in Internet Discussion ==----- http://www.dejanews.com/rg_mkgrp.xp Create Your Own Free Member Forum Received on Wed Aug 05 1998 - 08:36:57 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US