Horton

Re: pg15b3: crash in paralell vacuum

Justin Pryzby <pryzby@telsasoft.com>

View thread View on postgr.es
From: Justin Pryzby <pryzby@telsasoft.com>
To: Masahiko Sawada <sawada.mshk@gmail.com>
Cc: pgsql-hackers@postgresql.org, Amit Kapila <akapila@postgresql.org>
Date: 2022-08-18T14:52:36Z
Lists: pgsql-hackers 
On Thu, Aug 18, 2022 at 11:24:22PM +0900, Masahiko Sawada wrote:
> The status = 11 is invalid value. Probably because indstats was not
> initialized to 0 as I mentioned.
> 
> Justin, if it's reproducible in your environment, could you please try
> it again with the attached patch?

Yes, this seems to resolve the problem.

Thanks,
-- 
Justin

Commits

  1. Initialize index stats during parallel VACUUM.

  2. Prevent access to no-longer-pinned buffer in heapam_tuple_lock().