On Mon, Feb 25, 2019 at 10:49 PM David Steele <david@pgmasters.net> wrote:
> Worse, they have scripted the deletion of backup_label so that the
> cluster will restart on crash. This is the recommendation from our
> documentation after all. If that script runs after a restore instead of
> a crash, then the cluster will be corrupt -- silently.
I've always thought that the biggest problem is the documentation.
While the "Continuous Archiving and Point-in-Time Recovery (PITR)"
section of the user docs isn't quite as bad as it used to be, it's
still fundamentally structured "bottom up". That's bad.
I think that it would improve matters enormously if that section of
the documentation, and possibly the entire chapter was restructured to
be "top down". This should include prominent links to specific third
party projects, including Barman, pgBackRest, and others. This whole
area has become much more contentious than it needs to be, and I fear
that we cannot see the forest for the trees.
That said, I don't think that we should be in a hurry to fully remove
the exclusive backup mode. Home-spun backup solutions are a bad idea,
but removing exclusive mode strikes me as corrective over-reaction,
and a technical solution to an organizational problem. There are
definitely people that use the interface in a sensible way, despite
its shortcomings.
--
Peter Geoghegan