Re: backend crash on DELETE, reproducible locally

Karsten Hilbert <karsten.hilbert@gmx.net>

From: Karsten Hilbert <Karsten.Hilbert@gmx.net>
To: pgsql-general@lists.postgresql.org
Date: 2018-11-01T15:41:13Z
Lists: pgsql-hackers, pgsql-general

Attachments

On Thu, Nov 01, 2018 at 12:27:23PM -0300, Alvaro Herrera wrote:

> Ah, now this is interesting.  Can you please supply the definition of
> the table?

Attached.

> I'm wondering if there is a partitioned table with an FK to
> this one.

There is. Both ref.auto_hint and clin.suppressed_hint are
using inheritance (from audit.audit_fields). However, GNUmed
does not use inheritance for explicit partitioning but rather
similar to how classes and subclasses are used in OO languages.

>  I'm not quite seeing how come 'tup' is NULL there.  Can you
> 'print trigdata' in frame 2?

Sure, how ? :-)

(I can surely type "print trigdata" but does that already
auto-select from "frame 2" ?)

Karsten
-- 
GPG  40BE 5B0E C98E 1713 AFA6  5BC0 3BEA AC80 7D4F C89B

Commits

  1. Disable recheck_on_update optimization to avoid crashes.

  2. Allow HOT updates for some expression indexes