Horton

Re: pg_upgrade and materialized views

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andres Freund <andres@anarazel.de>
Cc: Claudio Freire <klaussfreire@gmail.com>, pgsql-bugs@postgresql.org
Date: 2018-02-20T22:29:01Z
Lists: pgsql-bugs 
Andres Freund <andres@anarazel.de> writes:
> I wonder if there's a scenario in which a schema restore uses enough
> xids to get close to anti-wraparound territory?

Interesting question ... you'd need one heck of a lot of objects in
the cluster, but we've certainly heard of people with lots of objects.

We could stave that problem off by running the restore steps in
--single-transaction mode, if it weren't that pg_restore will reject the
combination of --create and --single-transaction.  I wonder if we should
allow that, specifying that it means that the single transaction starts
after we reconnect to the new DB.

			regards, tom lane

Commits

  1. Repair pg_upgrade's failure to preserve relfrozenxid for matviews.