Re: BUG #17245: Index corruption involving deduplicated entries - Mailing list pgsql-bugs

From Pavel Borisov
Subject Re: BUG #17245: Index corruption involving deduplicated entries
Date
Msg-id CALT9ZEFr2B5yFw==1TfYZY-hC-6JvkN3MeUsZ1NkrDDuGfjQ9A@mail.gmail.com
Whole thread Raw
In response to Re: BUG #17245: Index corruption involving deduplicated entries  (Andres Freund <andres@anarazel.de>)
Responses Re: BUG #17245: Index corruption involving deduplicated entries  (Peter Geoghegan <pg@bowt.ie>)
List pgsql-bugs
One hypothesis just out-of-the air:
If heap page visibility map bit is wrongly set to all_visible, index-only scan can give duplicates (in effect new and old versions of the same row), with no apparent corruption of neither index nor heap relation. It just skips checking visibility of tuples in heap. Not sure it's the case now, but I've seen this previousy.

It could be diagnosed by deletion of VM file for a relation, the error (if it is this error) will disappear.

--
Best regards,
Pavel Borisov

Postgres Professional: http://postgrespro.com

pgsql-bugs by date:

Previous
From: Thomas Munro
Date:
Subject: Re: BUG #17245: Index corruption involving deduplicated entries
Next
From: Andres Freund
Date:
Subject: Re: BUG #17245: Index corruption involving deduplicated entries