On Mon, Apr 06, 2020 at 11:03:20AM +0300, Andrus wrote:
> If slave stops responing, master server will create files in pg_wal directory.
> If disk becomes full, master server crashes also.
>
> How to avoid this ?
>
> If disk is nearly full, master should stop additional files creation
> (and maybe stop or disable replication slot).
> Postgres 12 in Debian 10 is used.
When you use replication slots, it is very important to put in place a
monitoring solution to check if too much WAL is retained, and note
that there is nothing able to do that natively in core Postgres.
There are however multiple ways to solve this problem, like a
background worker (for the slot monitoring as well as optionally
killing and/or dropping), a simple cron job or even check_postgres.
--
Michael