Re: Problem with pg_upgrade?

Bruce Momjian <bruce@momjian.us>

From: Bruce Momjian <bruce@momjian.us>
To: Heikki Linnakangas <heikki.linnakangas@enterprisedb.com>
Cc: Robert Haas <robertmhaas@gmail.com>, Alvaro Herrera <alvherre@commandprompt.com>, Jeff Davis <pgsql@j-davis.com>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2011-03-31T16:07:32Z
Lists: pgsql-hackers
Heikki Linnakangas wrote:
> On 31.03.2011 17:55, Bruce Momjian wrote:
> > I will work on code to allow autovacuum_max_workers to be set to zero in
> > HEAD and 9.0, and have pg_upgrade us that.
> 
> We've intentionally not allowed the user to disable anti-wraparound 
> autovacuum before. Do we really want to allow it now for the sake of 
> pg_upgrade?

Not sure.

> >  I think the maintenance
> > overhead of an invisible variable is too much.
> 
> A simple GUC or command-line switch isn't much code.

Well, is this going to show in SHOW ALL or pg_settings?  Do we have the
ability to easily disable display of this?

> Is the problem just that the clog files get removed too early, or is 
> there something else? If it's just the clog files, we could simply copy 
> them (again) after updating datfrozenxids.

The problem is that pg_upgrade through pg_dumpall is setting
pg_database/pg_class frozen xid values and I can't have autovacuum
modifying the system while this is happening.

-- 
  Bruce Momjian  <bruce@momjian.us>        http://momjian.us
  EnterpriseDB                             http://enterprisedb.com

  + It's impossible for everything to be true. +