From: Tom Lane <tgl@sss.pgh.pa.us>
To: Rick Otten <rottenwindfish@gmail.com>
cc: Pietro Pugni <pietro.pugni@gmail.com>, pgsql-performance@postgresql.org
Subject: Re: Disk filled-up issue after a lot of inserts and drop schema
In-reply-to: 
 <CAMAYy4K6nwkqqdRcWR4cNeYvCXAH7h4WAo6vq1AkX3+MYOezsw@mail.gmail.com>
References: <AEA46136-0A29-4F3C-8966-90178EE44B4B@gmail.com>
 <CAMAYy4K6nwkqqdRcWR4cNeYvCXAH7h4WAo6vq1AkX3+MYOezsw@mail.gmail.com>
Comments: In-reply-to Rick Otten <rottenwindfish@gmail.com>
	message dated "Wed, 14 Sep 2016 10:17:41 -0400"
Date: Wed, 14 Sep 2016 10:44:39 -0400
Message-ID: <28460.1473864279@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-performance-owner@postgresql.org

Rick Otten <rottenwindfish@gmail.com> writes:
> I don't know why something still has an open file descriptor on something
> you believe has been removed, but at least that explains why you are
> experiencing the discrepancy between "du" and the real available space on
> the disk.

Yeah, the reported behavior clearly indicates that some PG process is
holding open files that should have been dropped (and were unlinked).
That's a bug, but there's not enough info here to find and fix it.

If we're really lucky, this is the same bug that Andres found and fixed
last week:
https://git.postgresql.org/gitweb/?p=postgresql.git;a=commitdiff;h=26ce63ce76f91eac7570fcb893321ed0233d62ff

but that guess is probably too optimistic, especially if it's a background
process (such as the checkpointer process) that is holding the open files.

If you can reproduce this, which I'm guessing you can, please use
"lsof" or similar tool to see which Postgres process is holding open
references to lots of no-longer-there files.

			regards, tom lane


-- 
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance