Date: Tue, 24 Mar 2020 12:42:59 -0700
From: Andres Freund <andres@anarazel.de>
To: Michael Paquier <michael@paquier.xyz>
Cc: Julien Rouhaud <rjuju123@gmail.com>, Justin King <kingpin867@gmail.com>,
	pgsql-general@lists.postgresql.org, kgrittn@gmail.com
Subject: Re: PG12 autovac issues
Message-ID: <20200324194259.pidupqxtqnquk4wu@alap3.anarazel.de>
References: <20200319235626.wz4b57hbhuuy4w7j@alap3.anarazel.de>
 <CAE39h23xR9F8qbo5vj9+KL1nFnnTaYJrFde0BtWczQy_+De39Q@mail.gmail.com>
 <20200320190317.gdxwzexdfutymqs6@alap3.anarazel.de>
 <20200323152247.GB52612@nol>
 <20200323162303.s7ay5hjdvimtkz6u@alap3.anarazel.de>
 <20200323194725.GC52612@nol>
 <20200323200051.qzyqvnips7o3ai4g@alap3.anarazel.de>
 <20200324052606.GE1836@paquier.xyz>
 <20200324054039.ulrxbzo2ycy7pu5i@alap3.anarazel.de>
 <20200324061238.GG1836@paquier.xyz>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20200324061238.GG1836@paquier.xyz>
Precedence: bulk

Hi,

On 2020-03-24 15:12:38 +0900, Michael Paquier wrote:
> > Well, there's no logging of autovacuum launchers that don't do anything
> > due to the "skipping redundant" logic, with normal log level. If somehow
> > the horizon logic of autovacuum workers gets out of whack with what
> > vacuumlazy.c does, then you'd not get any vacuums. But a usage level
> > triggered analyze could still happen on such a table, I think.
> 
> What surprised me the most is that the same table happened to be
> analyzed again and again after the launcher began its blackout.

Well, if there's an issue with the "redundant" logic, that would be a
not too surprising outcome. It's quite plausible that one or two tables
in the database would get enough changes to occasionally need to be
analyzed. If the workload is steady, that could e.g. work out to every
~17 minutes. All tables that autovacuum things are not wraparound
threatened will be skipped, but ones that are will get both vacuum and
analyze queued. The redundant logic could then entirely skip all
vacuums - but there's no equivalent for analyze.

> > While looking at this issue I found a few problems, btw. That seems more
> > like a -hackers discussion, so I started:
> > https://postgr.es/m/20200323235036.6pje6usrjjx22zv3%40alap3.anarazel.de
> 
> Yes, let's discuss there.

Cool. Would also be good if you could expand on the thread introducing
the "redundant" logic.

Greetings,

Andres Freund