Chris,

There was a bug with the pre-spawn option on release 7.3.2 which caused
system halt (I went into that bug too...). There is a patch available from
Oracle for that bug. Is 7.3.2 the version you're using?

Also, the implementation of SQL*Net's MTS configuration on MPE was not quite
stable (well, not with release 7.2.3). We configured MTS at first when we
started using Oracle, but we gave up after a while because of intermittent
lost of connection. Perhaps it is more stable now with recent release, but
we didn't try it again.

If you're not in production yet with your sytem, I suggest to use the latest
version of Oracle for MPE, which is 7.3.4.4. This version is supported for
both 5.5 and 6.0 (I think). And since it's their latest on MPE, you get
easier support from Oracle.

Michel Gauthier
[log in to unmask]

chris bullivant <[log in to unmask]> a écrit dans le message :
39164C7E.83F0909E@usq.edu.au...
> Hello anyone,
>         I come from a UNIX background, and although I've been *using*
> MPE for about 2 years, when it comes to anything of a deeper system
> nature I'm at a bit of a loss.
>
> For about 6 months (on and off) I've been working on converting an
> Interbase database to an ORACLE database. This is because our front end
> (screens, reports etc.) are Powerhouse and COGNOS no longer support
> Interbase.
>
> The problem I'm getting *seems* to be: when I test the load by
> simulating about 20 users trying to load the main menu screen
> simultaneously, it works (verry slowly) for a short while then the
> various sessions get TNS errors along the lines of: 00513 : Destination
> Host unreachable; 12203 : Unable to connect to destination; 12564 :
> TNS:connection refused; and a few others. The last time I killed the
> LISTENER process. On one occasion I got the error when I was the only
> person hitting the database.
>
> I thought that the Listener didn't have enough handlers, so I copied the
> bit on Pre-Spawning out of the documentation, and stopped, then started
> the Listener. Shortly afterwards the entire machine crashed (I was not
> at all popular). The System Manager believes that if you change the
> Listener parameters, you have to re-boot the machine (can this possibly
> be right??).
>
> Does this scenario sound in any way familiar to anyone? I would
> appreciate any insight that anyone has. I'm very much struggling in the
> dark here, so *anything* would help.
>
> Thanks in advance,
>
> Chris Bullivant,
> University of Southern Queensland,
> Australia
> [log in to unmask]
>