On Sun, Aug 26, 2012 at 10:25 PM, David Pinkerton
<david.pinkerton@betteroff.net> wrote:
>
> First let me prefix this with "I'm not a DBA, I am a Linux Sys Admin"
>
> Could someone provide a simple overview for me.
>
> We have a postgres 9.0 server which is about to run out of diskspace. I
> have built a new server and and set up a play virtual server so I can
> experiment with pgpool.
> I can create a db on the new server and populate it. Data appears on the
> virtual.
>
> So far so good.
>
> My question(s)...
>
> 1) how do I force the replication an existing database? (preference is zero
> downtime)
> 2) where do i install pg-pool if I want to rebuild the original server and
> then have load balancing across the 2 physical servers?
> 3) do i need to change all my scripts to talk on the pg-pool (ie. 9999 )
> port instead of postgres (5432) or do I make postgres listen on loopback and
> have pg-pool bind to the external ip (assuming its installed on the same
> server)?
>
>
>
> The documents on the wiki seem to jump from the one line descriptions of
> what the software can do, straight into configuration.
>
> Is there a scenario documents anywhere, that can show pro's and con's of
> each setup mode and is written in non-dba speak.
For emergency low-downtime switch of database from one server to
another, I'm wondering why you didn't explore the PITR route --
capturing a file system backup and grabbing the logs. You can go even
further and do HS/SR for the replica. pgpool (which i'm not very
familiar with) might also be a great tool for doing that -- I'm just
wondering what your thought process was.
merlin