From: Tom Lane <tgl@sss.pgh.pa.us>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>
cc: Michael Paquier <michael@paquier.xyz>, pgsql-hackers@lists.postgresql.org
Subject: Re: Odd, intermittent failure in contrib/pageinspect
In-reply-to: <20210118224005.GA3146@alvherre.pgsql>
References: <20210118224005.GA3146@alvherre.pgsql>
Comments: In-reply-to Alvaro Herrera <alvherre@alvh.no-ip.org>
	message dated "Mon, 18 Jan 2021 19:40:05 -0300"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <416875.1611010060.1@sss.pgh.pa.us>
Date: Mon, 18 Jan 2021 17:47:40 -0500
Message-ID: <416876.1611010060@sss.pgh.pa.us>
Precedence: bulk

Alvaro Herrera <alvherre@alvh.no-ip.org> writes:
> On 2021-Jan-18, Tom Lane wrote:
>> [ thinks for a bit... ]  Does the checkpointer pin pages it's writing
>> out?  I guess it'd have to ...

> It does, per SyncOneBuffer(), called from BufferSync(), called from
> CheckPointBuffers().

Right, then we don't need any strange theories about autovacuum,
just bad timing luck.  whelk does seem pretty slow, so it's not
much of a stretch to imagine that it's more susceptible to this
corner case than faster machines.

So, do we have any other tests that are invoking a manual vacuum
and assuming it won't skip any pages?  By this theory, they'd
all be failures waiting to happen.

			regards, tom lane