Message-ID: <1512063321.2328.20.camel@cybertec.at>
Subject: Re: PG 10 streaming replication pg_wal question
From: Laurenz Albe <laurenz.albe@cybertec.at>
To: ghiureai <isabella.ghiurea@nrc-cnrc.gc.ca>,
	pgsql-admin@lists.postgresql.org
In-Reply-To: <5A203DC7.40509@nrc-cnrc.gc.ca>
References: <5A203DC7.40509@nrc-cnrc.gc.ca>
Organization: Cybertec
Content-Type: text/plain; charset="UTF-8"
Date: Thu, 30 Nov 2017 18:35:21 +0100
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Precedence: bulk

ghiureai wrote:
> I am testing PG10 streaming replication ,  with archiving off, I have 
> pg_wal on separate directory  to monitor the growth , I had the slave 
> offline yesterday  for  more than 20h while I was restoring one of db 
> (60GB) with pg_restore from a backup taken with pg_dump, today I brought 
> the slave PG host online and replication catch up  nicely , master and 
> slave are in sync now, but  on master host the pg_wal is still same  
> large size ( as yesterday)  even after all the wal files had been 
> applied to salve, would replication process not suppose to removed the 
> wal files  on master after being applied to salve ?
> ( do I need to manually implement a cleanup job of this wal files, as 
> mentioned archiving is off on  both servers)

Never remove WAL files yourself.

pg_wal will shrink eventually.
At the next checkpoint, PostgreSQL will remove all WAL files
that were completed and archived successfully since the
previous checkpoint, thus reducing WAL size a little.

If there is activity on the databases, pg_wal will eventually
shrink back to max_wal_size.

Yours,
Laurenz Albe