On Mon, Sep 11, 2017 at 7:02 AM, Dave Florek <dave.a.florek@xxxxxxxxx> wrote:
Hi,I'm trying to setup streaming replication on a cluster of postgresql databases and I followed the instructions outlined here (https://wiki.postgresql.org/wiki/Streaming_Replication ) but I'm stuck on the archive and restore commands. In reading the comments from the archive command, WAL archiving needs a shared directory between the primary and standby machines in order to replicate. Is this something actually required? If so, does postgresql have any recommendations on where this shared directory ought to be hosted?
You can use streaming replication without any WAL archive at all. If the only reason you want a WAL archive is so that the standby can fall behind the primary without losing the ability to catch back up again, then you can use a replication slot to make the primary hold onto WAL files until the standby is done with them. That is going to be simpler than using a WAL archive, provided the master has enough room to keep them around should the standby fail for a while.
If you want a WAL archive for other reasons, then the choice of how to host it would depend on those reasons.
Cheers,
Jeff
