Am 11.05.2017 um 15:58 schrieb Martin Goodson:
Hello. First time poster, so please be gentle :)
I have a PostgreSQL 9.6 database cluster running on a standalone
Redhat 7.2 (Maipo) server.
This may seem like a silly question. It probably *is* a silly
question, so apologies in advance.
PostgreSQL was installed using the "EDB Postgres Standard" installer
from EnterpriseDB -
https://www.enterprisedb.com/software-downloads-postgres - and
installed to a specific location - /db_<dbname>. No repositories,
nothing in a 'standard' place.
I have now been tasked with putting repmgr onto this box and
converting it from a standalone to a multiple node setup (I have two
additional servers lurking in the background ready for use :) ).
Numerous articles on repmgr all seem to suggest it should be a
relatively simply operation once the software is in place. Register
the master. Clone standbys. Ought to be *relatively* straight-forward.
Unfortunately, I'm having problems at the first hurdle - putting the
software in place. We initially tried installing from the postgres
repository rpm - which insisted, of course, on installing PostgreSQL
as well as repmgr in /usr/pgsql-9.6. Obviously not the right place for
it given we're running our binaries from
/db_dbname/app/postgres/9.6.2-3/bin
Is there a simple way to use repmgr from the package, perhaps by
copying/linking files from that location to ours? (e.g.
copying/linking some files from lib, bin, contrib, etc).
I think, repmgr *should* work with EDB-Version of PostgreSQL. You can
set pg_bindir in repmgr.conf. I would also suggest create symlinks in
/etc to the repmgr.conf.
Btw.: you also set
|service_start_command service_stop_command service_restart_command
service_reload_command service_promote_command|
Keep me informed if that works, thx.
Andreas
--
2ndQuadrant - The PostgreSQL Support Company.
www.2ndQuadrant.com