Re: [GENERAL] Multiple Postgres Instances

It looks like the .profile and /etc/profile were overriding the -D option.
I set the profile to this:

PATH=/sbin:/usr/sbin:/bin:/usr/bin:/usr/local/bin:/usr/local/pgsql/bin
MANPATH=:/usr/local/pgsql/bin
PGLIB=/usr/local/pgsql/lib
# note: PGDATA overwrites the -D startup option
# PGDATA=/usr/local/pgsql/data

I also had to take the PGDATA value out of the /etc/profile (I had it set
there because most of the users on the system are database users). Then
changed the rc script to this:

[ -x /usr/local/pgsql/bin/postmaster ] && {
        su -l postgres -c '/usr/local/pgsql/bin/postmaster -p5432 -D
/usr/local/pgsql/data_5432 -S -i'
        echo -n ' pgsql 5432'
}
[ -x /usr/local/pgsql/bin/postmaster ] && {
        su -l postgres -c '/usr/local/pgsql/bin/postmaster -p5433 -D
/usr/local/pgsql/data_5433 -S -i'
        echo -n ' pgsql 5433'
}

Presto! Everything seems to be working! Thanks for the suggestions!

David


On Thu, 21 Jan 1999, Jeff Hoffmann wrote:

> David Giffin wrote:
> >
> > Hello,
> >
> > I'm trying to run postgres on two different post 5432 and port 5433 with
> > different data directories for each instance:
> >
> > [ -x /usr/local/pgsql/bin/postmaster ] && {
> >         su -l postgres -c '/usr/local/pgsql/bin/postmaster -S -i'
> >         echo -n ' pgsql'
> > }
> >
> > [ -x /usr/local/pgsql/bin/postmaster ] && {
> >         su -l postgres -c '/usr/local/pgsql/bin/postmaster
> > -D/usr/local/pgsql/data_5433 -p5433 -S -i'
> >         echo -n ' pgsql'
> > }
> >
> > But when I do a psql -p 5433 and drop a table it seems to drop in both
> > instances. If anyone has attempted this before I like to hear how it
> > worked.
> >
> > David
>
> just a guess on my part, but i think what you're going to want to do is
> set the PG_DATA environment variable to /usr/local/pgsql/data_5433 in
> your script before you start the second instance.  they screwed around
> with the ability to specify absolute paths to the data directory in
> 6.4.x and didn't document it, so i'm not sure what the truth is about
> it.  i'm not sure if that's what  is causing the problem in _this case_,
> but it has happened with other cases.  (it makes good sense to require
> you to use environment variables in all cases, as far as i'm
> concerned.)  i submitted some doc patches, but apparently they haven't
> shown up yet.
>
> again, this is just a guess, but an educated one.
>
> jeff
>
> jeff@remapcorp.com
>