Horton

Re: More race conditions in logical replication

Michael Paquier <michael.paquier@gmail.com>

View thread View on postgr.es
From: Michael Paquier <michael.paquier@gmail.com>
To: Alvaro Herrera <alvherre@2ndquadrant.com>
Cc: Robert Haas <robertmhaas@gmail.com>, Petr Jelinek <petr.jelinek@2ndquadrant.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>, Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
Date: 2017-08-10T07:22:23Z
Lists: pgsql-hackers

Attachments

On Tue, Aug 8, 2017 at 8:11 PM, Alvaro Herrera <alvherre@2ndquadrant.com> wrote:
> Robert Haas wrote:
>> On Mon, Aug 7, 2017 at 8:14 PM, Alvaro Herrera <alvherre@2ndquadrant.com> wrote:
>> > BTW, I noticed that the PG_WAIT_LOCK value that we're using for wait
>> > event here (and in the replication slot case) is bogus.  We probably
>> > need something new here.
>>
>> Yeah, if you're adding a new wait point, you should add document a new
>> constant in the appropriate section, probably something under
>> PG_WAIT_IPC in this case.
>
> Here's a patch.  It turned to be a bit larger than I initially expected.

Álvaro, 030273b7 did not get things completely right. A couple of wait
events have been added in the docs for sections Client, IPC and
Activity but it forgot to update the counters for morerows . Attached
is a patch to fix all that.
-- 
Michael

Commits

  1. Fix vertical spanning in table "wait_event Description".

  2. Fix replication origin-related race conditions

  3. Fix inadequacies in recently added wait events

  4. Fix race conditions in replication slot operations