On 2/18/18 7:34 PM, Don Seiler wrote:
> Looking to use pgBackRest to take a backup from a hot standby. I'm
> reading that pgBackRest still needs to connect to the primary and copy
> some files. My questions are:
> 1. What files does it need to copy? Config files? WAL files?
It copies files that are not replicated from the primary so that a
primary-style backup is created. Anything that is replicated (which is
by far the bulk of the data) is copied from the standby.
> 2. How does it connect? SSH?
> 3. Does pgBackRest need to be installed and configured on the primary
> as well?
Yes. Anyway, it's best to archive from the primary so a replication
failure does not affect your archiving.
Configuring pgBackRest, SSH, standby, and backup from standby are all
covered in the user guide.