Re: Further pg_upgrade analysis for many tables

On 23 November 2012 22:34, Jeff Janes <jeff.janes@gmail.com> wrote:

> I got rid of need_eoxact_work entirely and replaced it with a short
> list that fulfills the functions of indicating that work is needed,
> and suggesting which rels might need that work.  There is no attempt
> to prevent duplicates, nor to remove invalidated entries from the
> list.   Invalid entries are skipped when the hash entry is not found,
> and processing is idempotent so duplicates are not a problem.
>
> Formally speaking, if MAX_EOXACT_LIST were 0, so that the list
> overflowed the first time it was accessed, then it would be identical
> to the current behavior or having only a flag.  So formally all I did
> was increase the max from 0 to 10.

...

> It is not obvious what value to set the MAX list size to.

A few questions, that may help you...

Why did you pick 10, when your create temp table example needs 110?

Why does the list not grow as needed?

-- Simon Riggs                   http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training & Services