Re: Potential G2-item cycles under serializable isolation

Thomas Munro <thomas.munro@gmail.com>

From: Thomas Munro <thomas.munro@gmail.com>

To: Kyle Kingsbury <aphyr@jepsen.io>

Cc: Peter Geoghegan <pg@bowt.ie>, PostgreSQL mailing lists <pgsql-bugs@lists.postgresql.org>

Date: 2020-06-01T04:19:16Z

Lists: pgsql-bugs

On Mon, Jun 1, 2020 at 4:05 PM Kyle Kingsbury <aphyr@jepsen.io> wrote:
> I'll also see about getting a version of these tests that doesn't involve ON CONFLICT, in case that's relevant.

That should be interesting.  I'm wondering if the read of the old
value in the UPDATE case is not done with appropriate predicate locks,
so we miss a graph edge?

Commits

Doc: Add references for SI and SSI.
- 4c897197d793 9.5.23 landed
- 7295ac9c6c71 9.6.19 landed
- e2aca0d02d97 10.14 landed
- 648e56d3984d 11.9 landed
- eb1286413c0a 12.4 landed
- 4701efa9f741 13.0 landed
- 1575fbf1ac3c 14.0 landed
Improve comments for [Heap]CheckForSerializableConflictOut().
- 44eff2841059 13.0 landed
- 7aa4fb592530 14.0 landed
Avoid update conflict out serialization anomalies.
- 5940ffb22131 14.0 landed
- 6df7105e5d50 13.0 landed
- e620a38c23b9 12.4 landed
- dd616f2ec1c5 11.9 landed
- 19a6e1bc3083 10.14 landed
- e81dc2b7e2c4 9.6.19 landed
- c05f6b354ffe 9.5.23 landed
Implement genuine serializable isolation level.
- dafaa3efb75c 9.1.0 cited