Re: Backup "Best Practices"

Why wouldn't the streaming replica salvage that?  Are they expected to fail together?  Is the NFS share onto which you want to store your basebackup and WAL also expected to fail together with them?

That's why I specified *total* failure. If only the primary dies, that's a simple cutover to the secondary, and not a total failure - no need to go to backups of any sort under that scenario :-) I'm thinking more along the lines of server room burns down, natural disaster, etc - something that causes a total failure of the db system, necessitating recovery from an offsite backup.

your base backups and WAL archive should be replicated offsite, then.

-- 
john r pierce, recycling bits in santa cruz