Re: Subscription code improvements

From: Peter Eisentraut <peter(dot)eisentraut(at)2ndquadrant(dot)com>
To: Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com>
Cc: Petr Jelinek <petr(dot)jelinek(at)2ndquadrant(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Subscription code improvements
Date: 2017-08-05 01:29:31
Message-ID: 4a774467-b7bf-7ec5-3d8f-5516d135f10d@2ndquadrant.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On 8/4/17 12:02, Masahiko Sawada wrote:
> To make ALTER SUBSCRIPTION REFRESH being transactional, I prefer
> Petr's proposal. Because it can make ALTER SUBSCRIPTION and DROP
> SUBSCRIPTION stop the table sync workers that are in progress of
> copying data. I'm not sure killing table sync workers in DDL commands
> would be acceptable but since it can free unnecessary slots of logical
> replication workers and replication slots I'd prefer this idea.

OK, I have committed the 0004 patch.

> However, even with this patch there is still an issue that NOTICE
> messages "removed subscription for table public.t1" can be appeared
> even if we rollback ALTER SUBSCRIPTION REFRESH command as I mentioned
> on earlier thread. Since I think this behaviour will confuse users who
> check server logs I'd like to deal with it, I don't have a good idea
> though.

Maybe we can just remove those messages?

We don't get messages when we create a subscription about which tables
are part of it. So why do we need such messages when we refresh a
subscription?

--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Robert Haas 2017-08-05 02:05:56 Re: A bug in mapping attributes in ATExecAttachPartition()
Previous Message Peter Geoghegan 2017-08-05 01:02:02 Re: [WIP] Zipfian distribution in pgbench