I don't work on Oracle so I will speak from my experience with MySQL. MySQL
servers time out after the 8 hour standard disconnect for inactivity (this
can be adjusted in your my.conf file). To compensate for this we now run our
own connect checks for a valid dbh handle before it goes it all the trouble
to make one along with Apache::DBI
We have not had any more issues with the database connects since we moved to
this method. We do a ping and validate the dbh handle rather then blindly
accessing the dbh handle since Apache::DBI will only validate the dbh handle
on a connect.
Aaron Johnson
Perrin Harkins wrote:
> On Thu, 31 Aug 2000 [EMAIL PROTECTED] wrote:
> > What I think is going on is that the script gets killed by Oracle for
> > being idle and tries to ping the connection, but the ping fails.
>
> It is supposed to reconnect when the ping fails. I've had problems
> getting reconnects to Oracle 8 working. The "solution" we ended up with
> was to make processes that can't reconnect send an error page and
> exit. New processes are able to connect. I'm not sure what causes this
> problem.
>
> Since your problem is caused by your processes being idle for too long,
> this may go away when you move out of testing mode into a public release.
> You might want to tweak your MinSpareServers settings so that you won't
> have lots of idle processes hanging around.
>
> > >Rebild your mod_perl with the EVERYTHING=1 flag. That will get rid
> > of the >above error message.
> >
> > So I have to re-install it? Is there anything I need to do when I
> > rebuild it? Or do I just need to reinstall mod_perl as it's done in
> > the documentation?
>
> Just rebuild it and re-install as it shows in the docs.
>
> - Perrin
