From: | Bruce Momjian <bruce(at)momjian(dot)us> |
---|---|
To: | PostgreSQL-development <pgsql-hackers(at)postgreSQL(dot)org> |
Subject: | pg_upgrade and visibility map files |
Date: | 2012-11-24 18:16:05 |
Message-ID: | 20121124181605.GC1594@momjian.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
I have found an odd mistake in pg_upgrade 9.2, though I don't think it
is a problem. Specifically, pg_upgrade will copy a visibility map (vm)
>1GB extent file from the old cluster to the new, even if the
crash-safety of the old vm file does not match the new server. For
example, 19323_vm would not be copied, but 19323_vm.1 would be.
I don't think this is a problem because having a >1GB vm file would
require the table to be 64TB (1GB * 8 * 8k), and our max table size is
32TB. I am not planning to fix this in 9.2; this code is restructured
in git head and does not have this problem.
--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com
+ It's impossible for everything to be true. +
From | Date | Subject | |
---|---|---|---|
Next Message | Bruce Momjian | 2012-11-24 18:20:14 | Re: Further pg_upgrade analysis for many tables |
Previous Message | Andres Freund | 2012-11-24 17:38:18 | Re: Bugs in CREATE/DROP INDEX CONCURRENTLY |