On 11/13/20 7:28 AM, Jeremy Wilson wrote:
On Nov 13, 2020, at 10:23 AM, Tom Lane <tgl@xxxxxxxxxxxxx> wrote:
Unless ... could it be that there is another PG server active on the
machine, whose cluster lacks a "template1" database? Seems unlikely,
but you might try confirming with "ps auxww | grep post" or the like.
This is a test environment so only the two databases are on it and neither is running. pg_upgrade checks if either is running.
When you manually run the pg_upgrade pg_ctl script the server starts but
you cannot connect to any database in it correct?
What does pg_upgrade_server.log show when you do above?
I also wonder if the cluster you're trying to upgrade from has a
"template1" database. Not sure if anyone has ever tested whether
pg_upgrade can cope with the lack of one.
It’s there:
$ psql -p 54320 template1
psql (13.0, server 9.5.23)
Type "help" for help.
template1=#
--
Adrian Klaver
adrian.klaver@xxxxxxxxxxx
