Re: pg_upgrade bug found!

Bruce Momjian wrote:
> Tom Lane wrote:
> > Josh Berkus <josh@agliodbs.com> writes:
> > >> -- It will not lock any tables but will generate I/O.
> > 
> > > add:
> > > IMPORTANT: Depending on the size and configuration of your database,
> > > this script may generate a lot of I/O and degrade database performance.
> > > Users should execute this script during a low traffic period and watch
> > > the database load.
> > 
> > It might be worth suggesting that people can adjust their vacuum delay
> > parameters to control the I/O load.
> 
> I talked to Tom about this and I am worried people will adjust it so it
> takes days to complete.  Is that a valid concern?  Does anyone have a
> suggested value?

Josh Berkus helped me get lots more details on a wiki page for this:
http://wiki.postgresql.org/wiki/20110408pg_upgrade_fix

I will reference the wiki in the email announcement.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + It's impossible for everything to be true. +