Hiya,
We're looking into ways to cluster our database servers and
shared storage. Here's the low-down on our servers... all
run RedHat 7.3 with a few patches and are dual PIII 1.4GHz,
2GB, mirrored internal 36, Qlogic 2310 HBA to a LUN on our
Hitachi. PostgreSQL 7.3.3 is installed from source and the
one that came with RH has been removed.
There are currently 3 of these servers and each are designed
to run 1 instance of PostgreSQL with 10-20 databases of varying
size from 5 to 20GB. I have PostgreSQL pretty well tweaked to
maximize memory so I'm hesitant to have a 2nd instance running
on the server in the case of a failover.
This is what I'm curious about.
Server A has it's databases under /pg/dataA and Server B has
it's databases under /pg/dataB both are different LUN's on our
SAN. Should Server A choke, Server B should fsck /pg/dataA
and mount it locally. At this point it needs to either
(1) start a new instance of PostgreSQL listening on 5432/tcp
but bound to a different IP
or
(2) the current PostgreSQL "imports" the databases under
/pg/dataA
Each have their good points and problematic issues.
Thoughts?
Greg
--
Greg Spiegelberg
Sr. Product Development Engineer
Cranel, Incorporated.
Phone: 614.318.4314
Fax: 614.431.8388
Email: gspiegelberg@Cranel.com
Cranel. Technology. Integrity. Focus.