> On 4 Mar 2021, at 08:24, Peter Eisentraut <peter(at)eisentraut(dot)org> wrote:
> I think the attached would be a sensible fix.
That's pretty much what I ended up with as well.
I was first looking at checking the version of the binary against the
major_version_str in ClusterInfo by passing down cluster instead, but that
turns into chicken-and-egg since we don't set the extracted version until we've
checked that pg_ctl exists and works.
Thanks!
--
Daniel Gustafsson https://vmware.com/