Horton

Re: Corrupted btree index on HEAD because of covering indexes

Peter Geoghegan <pg@bowt.ie>

View thread View on postgr.es
From: Peter Geoghegan <pg@bowt.ie>
To: Teodor Sigaev <teodor@sigaev.ru>
Cc: Michael Paquier <michael@paquier.xyz>, Postgres hackers <pgsql-hackers@postgresql.org>, Anastasia Lubennikova <a.lubennikova@postgrespro.ru>
Date: 2018-04-24T17:12:28Z
Lists: pgsql-hackers

Attachments

On Tue, Apr 24, 2018 at 9:06 AM, Teodor Sigaev <teodor@sigaev.ru> wrote:
> Perfect!

Thanks!

> I would like to commit it but have some suggestions:

I attach a revised version, which has changes based on your feedback.

> to improve test stability it would be better to disable autovacuum:
> ALTER TABLE bttest_multi SET (autovacuum_enabled = false)

Done.

> 2) Pls, add to test DELETE as it done in
> 6db4b49986be3fe59a1f6ba6fabf9852864efc3e

Done. I will leave it to you to decide whether or not the original
create_index.sql test can now be removed.

> 3) It's not directly connected to this patch, but allocation of
> BtreeCheckState is not needed, it could be allocated on stack.
>
> 4) Nevertheless, I suggest to use palloc0 (or memset(0)) for
> BtreeCheckState. Now several fields of that structure could be not inited.

I like the idea of using palloc0(). Done that way.

-- 
Peter Geoghegan

Commits

  1. Add missing and dangling downlink checks to amcheck

  2. Fix wrong validation of top-parent pointer during page deletion in Btree.

  3. Add integrity-checking functions to pg_visibility.

  4. Make the handling of interrupted B-tree page splits more robust.