Re: improving wraparound behavior

Hi,

On 2019-05-03 22:11:08 -0400, Stephen Frost wrote:
> * Andres Freund (andres(at)anarazel(dot)de) wrote:
> > I don't think we necessarily need a new WAL record for what I'm
> > describing above (as XLOG_SMGR_TRUNCATE already carries information
> > about which forks are truncated, we could just have it acquire the
> > exclusive lock), and I don't think we'd need a ton of code for eliding
> > the WAL logged lock either. Think the issue with backpatching would be
> > that we can't remove the logged lock, without creating hazards for
> > standbys running older versions of postgres.
>
> While it's pretty rare, I don't believe this would be the only case of
> "you need to upgrade your replicas before your primary" due to changes
> in WAL. Of course, we need to make sure that we actually figure out
> that the WAL being sent is something that the replica doesn't know how
> to properly handle because it's from a newer primary; we can't simply do
> the wrong thing in that case.

Don't think this is severe enough to warrant a step like this. I think
for the back-patch case we could live with
a) move truncation to after the rest of vacuum
b) don't truncate if it'd error out anyway, but log an error

Greetings,

Andres Freund