Date: Tue, 23 Jul 2019 09:33:00 -0500
From: Kenneth Marshall <ktm@rice.edu>
To: Kristian Ejvind <Kristian.Ejvind@resurs.se>
Cc: "pgsql-performance@lists.postgresql.org"
 <pgsql-performance@lists.postgresql.org>
Subject: Re: zabbix on postgresql - very slow delete of events
Message-ID: <20190723143300.GG21184@aart.rice.edu>
References: <3EC00A37-57CF-4B43-A737-968A87B02641@contoso.com>
 <20190723125803.GD21184@aart.rice.edu>
 <FECD4549-3F9B-43E5-A475-3996D5E3F22F@resurs.se>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <FECD4549-3F9B-43E5-A475-3996D5E3F22F@resurs.se>
User-Agent: Mutt/1.5.21 (2010-09-15)
Precedence: bulk

On Tue, Jul 23, 2019 at 01:41:53PM +0000, Kristian Ejvind wrote:
> Thanks Kenneth. In fact we've already partitioned the largest history* and trends* tables
> and that has been running fine for a year. Performance was vastly improved. But since you
> can't have a unique index on a partitioned table in postgres 10, we haven't worked on that.
> 
> Regards
> Kristian

Hi Kristian,

Why are you not partitioning the events and alerts tables as well? That
would eliminate this problem and you already have the infrastructure in
place to support the management since you are using it for the history
and trends tables.

Regards,
Ken