| From: | Peter Geoghegan <pg(at)heroku(dot)com> |
|---|---|
| To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
| Cc: | Ivan Frolkov <ifrol2001(at)mail(dot)ru>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: UPSERT strange behavior |
| Date: | 2016-08-25 20:04:12 |
| Message-ID: | CAM3SWZRWhN=aYidDDcYffbUXAoRYZ0zD4LH5ovLN_do1BiSzJg@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Thu, Aug 25, 2016 at 12:59 PM, Peter Geoghegan <pg(at)heroku(dot)com> wrote:
> Maybe we should change the ordering of those IndexInfo structs to
> something more suitable, but it must be immutable (it cannot hinge
> upon the details of one particular DML statement).
I meant that it must be stable (not immutable), in the specific sense
that it cannot change without an ALTER TABLE or similar. The order of
insertion should be consistent among all backends that might be
performing DML against the table at any given time, since, in general,
to do otherwise risks deadlocking (perhaps only with hash indexes, or
a much earlier version of the nbtree am that still used heavyweight
locks).
--
Peter Geoghegan
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Alvaro Herrera | 2016-08-25 20:05:09 | Re: increasing the default WAL segment size |
| Previous Message | Peter Geoghegan | 2016-08-25 19:59:41 | Re: UPSERT strange behavior |