Horton

Re: Old row version in hot chain become visible after a freeze

Michael Paquier <michael.paquier@gmail.com>

View thread View on postgr.es
From: Michael Paquier <michael.paquier@gmail.com>
To: Peter Geoghegan <pg@bowt.ie>
Cc: "Wood, Dan" <hexpert@amazon.com>, "pgsql-bugs@postgresql.org" <pgsql-bugs@postgresql.org>
Date: 2017-09-01T03:46:44Z
Lists: pgsql-bugs 
On Fri, Sep 1, 2017 at 12:25 PM, Peter Geoghegan <pg@bowt.ie> wrote:
> On Thu, Aug 31, 2017 at 3:36 PM, Wood, Dan <hexpert@amazon.com> wrote:
>> I’ve found a bug in Postgres which causes old row versions to appear in a
>> table.  DEAD rows in a hot chain are getting frozen and becoming visible.
>> I’ve repro’d this in both 9.6.1 and 11-devel.
>
> I can confirm that this goes all the way back to 9.3, where "for key
> share"/foreign key locks were introduced.

Hm. That looks pretty bad to me... It is bad luck that this report
happens just after the last round of minor releases has been out, and
we would have needed at least a couple of days and a couple of pairs
of eyes to come up with a correct patch. (I haven't looked at the
proposed solution and the attached patch yet, so no opinions yet).
-- 
Michael

Commits

  1. Fix freezing of a dead HOT-updated tuple

  2. Fix potential data corruption during freeze

  3. Clarify the contract of partition_rbound_cmp().