From owner-freebsd-fs@FreeBSD.ORG Mon Oct 3 18:50:23 2011 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id A70EB106566C; Mon, 3 Oct 2011 18:50:23 +0000 (UTC) (envelope-from artemb@gmail.com) Received: from mail-yx0-f182.google.com (mail-yx0-f182.google.com [209.85.213.182]) by mx1.freebsd.org (Postfix) with ESMTP id 3A8DC8FC08; Mon, 3 Oct 2011 18:50:22 +0000 (UTC) Received: by yxk36 with SMTP id 36so4806688yxk.13 for ; Mon, 03 Oct 2011 11:50:22 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=mime-version:sender:in-reply-to:references:date :x-google-sender-auth:message-id:subject:from:to:cc:content-type; bh=WK6EHV9rfPYThkhy2JZZhGH/TMrDe2Kv/YHgmZf2Ovg=; b=HRwAFJwEue+nn5UVt2XM+V5alRYp8x/VCJ3O6fK8SYfFzm/HmUn27HPvMpWqYc3hv+ JZJd8/xFKpjPMyPHFeOBgN0W3ZVVvZ8PEkRYMinHOL5ezATYBG1J8bv3lw2on3O3iYep gZF/rpWxmAkaPjwBWGjiwl9OOQIEuWxxcYX9Y= MIME-Version: 1.0 Received: by 10.236.185.37 with SMTP id t25mr1513595yhm.131.1317667822440; Mon, 03 Oct 2011 11:50:22 -0700 (PDT) Sender: artemb@gmail.com Received: by 10.236.103.33 with HTTP; Mon, 3 Oct 2011 11:50:22 -0700 (PDT) In-Reply-To: <4E8A0013.4070008@fsn.hu> References: <20111002020231.GA70864@icarus.home.lan> <4E899C8E.7040305@fsn.hu> <4E8A0013.4070008@fsn.hu> Date: Mon, 3 Oct 2011 11:50:22 -0700 X-Google-Sender-Auth: HzRjBDkGPpOTJx_3decdyz5HLHU Message-ID: From: Artem Belevich To: Attila Nagy Content-Type: text/plain; charset=ISO-8859-1 Cc: freebsd-fs@freebsd.org, Adrian Chadd , delphij@freebsd.org Subject: Re: is TMPFS still highly experimental? X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 03 Oct 2011 18:50:23 -0000 On Mon, Oct 3, 2011 at 11:33 AM, Attila Nagy wrote: > Sounds plausible. BTW, it may be possible that the ARC limits are not needed > anymore, they are here from the times, where on a 64 GB machine ARC hovered > around 2-5 GBs without setting these (arc_min was even higher then). You do need tuning. FreeBSD defaults are way too conservative for ZFS. While they may get you a working ZFS config, you will end up with ARC using only small fraction of your memory, which is probably not the best setup for a filesrever. > BTW, the user space programs fit into around 1-2 GB RAM on this machine > typically. Well, most of the time. :) Yup. 2GB for apps, 2-5GB for ZFS ARC and you end up with 50+GB of memory that's effectively wasted. The trick is in finding the balance where ZFS uses as much memory as you can give it without causing trouble for other subsystems. ARC back-pressure mechanism works somewhat better now than it used to, but it's still very far from perfect. Ideally it would be nice to integrate ARC with the FreeBSD's unified cache, but that's not a trivial task, as far as I can tell. --Artem >> >> I'd start with doubling kmem_size and, possibly, reducing arc_max to >> the point where it stops putting pressure on tmpfs. >> > I know there are several differences, but it would be very good to have > similar behaviour with UFS. I guess it's quite evident that tmpfs can eat > the file system cache, and I know it may be not so trivial to solve this > with ZFS. :) > > Will try it, thanks. >