From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from magic.merlins.org ([209.81.13.136]:42358 "EHLO mail1.merlins.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751554AbaGDPHV (ORCPT ); Fri, 4 Jul 2014 11:07:21 -0400 Date: Fri, 4 Jul 2014 08:07:10 -0700 From: Marc MERLIN To: Russell Coker Cc: Satoru Takeuchi , linux-btrfs@vger.kernel.org Message-ID: <20140704150710.GM26932@merlins.org> References: <20140704011938.GO11539@merlins.org> <53B6486D.9010006@jp.fujitsu.com> <20140704142416.GI26932@merlins.org> <3262371.bchVJ2xl3c@xev> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <3262371.bchVJ2xl3c@xev> Subject: Re: Is btrfs related to OOM death problems on my 8GB server with both 3.15.1 and 3.14? Sender: linux-btrfs-owner@vger.kernel.org List-ID: On Sat, Jul 05, 2014 at 12:45:55AM +1000, Russell Coker wrote: > > But the last times I had this OOM problem with 3.15.1 it was happening > > within 6 hours sometimes, and I was not starting scrub every time the > > system booted, so scrub may be partially responsible but it's not the > > core problem. > > It would be a good idea to run a few scrubs and see if this is a repeatable > problem. If it's a repeatable problem then it's something to fix regardless > of whether it's the only issue you have. Fair point. Re-running scrub now. That will take 36H or so :) > If a scrub can reliably trigger the problem it would be good to test 3.14 for > the same behavior. Knowing whether it's a regression would help the > developers. I'm already back to 3.14. 3.15 was dying about once a day without scrub, enough that this was causing me real problems (it's a server that is supposed to do work :) ). > Even without much swap 8G should be a plenty. My main workstation has 4G of > RAM and 6G of swap. I almost never use more than 3G of swap because the > system becomes so slow as to be almost unusable when swap gets to 4G (Chromium > is to blame). However that is for a 120G non-RAID filesystem. Presumably a > RAID array will need some more kernel memory and a larger filesystem will also > need a little more, but it still shouldn't be that much. You're correct. I only have more swap than ram, both as a habit in case I ever want to hibernate a system (I don't with this one) and in case some userland stuff leaks a lot (I've had versions of Xorg leak and that would improve the time between which I had to restart X and lose all my window state :) ). Marc -- "A mouse is a device used to point at the xterm you want to type in" - A.S.R. Microsoft is to operating systems .... .... what McDonalds is to gourmet cooking Home page: http://marc.merlins.org/ | PGP 1024R/763BE901