Date: Thu, 19 Sep 2013 00:08:08 +0800 (SGT) From: Patrick Dung <patrick_dkt@yahoo.com.hk> To: "freebsd-hackers@freebsd.org" <freebsd-hackers@freebsd.org> Cc: "ivoras@freebsd.org" <ivoras@freebsd.org> Subject: About Transparent Superpages and Non-transparent superapges Message-ID: <1379520488.49964.YahooMailNeo@web193502.mail.sg3.yahoo.com> In-Reply-To: <mailman.2681.1379448875.363.freebsd-hackers@freebsd.org> References: <mailman.2681.1379448875.363.freebsd-hackers@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
>Date: Tue, 17 Sep 2013 18:06:32 +0200=0A>From: Ivan Voras <ivoras@freebsd.= org>=0A>To: freebsd-hackers@freebsd.org=0A>Subject: Re: About Transparent S= uperpages and Non-transparent=0A>=A0=A0=A0 superapges=0A>Message-ID: <l19um= 0$jvd$1@ger.gmane.org>=0A>Content-Type: text/plain; charset=3D"utf-8"=0A>= =0A>On 17/09/2013 17:01, Patrick Dung wrote:=0A>>=0A>>=0A>> Hello,=0A>>=0A>= >=0A>> I have posted the question in freebsd-questions but have not get fee= dback, so I tried to asked in here.=0A>>=0A>> 1.=0A>> Transparent Superpage= s was in FreeBSD for a few years.=0A>> I would like to know if there is any= benchmark or real world performance experience about this setting.=0A>>=0A= >> 2.=0A>> I have seen somewhere that non-transparent superpages was being = developed in HEAD too.=0A>> Any insight on it? Please correct me if it is n= ot the case.=0A>=0A>By "non-transparent" do you mean explicit huge pages AP= I which allows=0A>them to be allocated on-demand rather than heuristically,= such as was=0A>implemented in Linux for at least 8 years (http://goo.gl/8q= ZX4D,=0A>https://lwn.net/Articles/375096/) and supported by major software= =0A>products (http://goo.gl/prxjjo, http://goo.gl/fQOLwO,=0A>http://goo.gl/= pr7Tbb, http://goo.gl/Y9qtWk, http://goo.gl/M0l7LL, etc.)?=0A>I haven't hea= rd about it (but I hope I'm wrong :) ).=0A>=0A=0AFor question no 2: Yes. oO= I should say superpages , large pages or huge page support.=0A=0ABack to m= y questions:=0A1.=0ATransparent Superpages was in FreeBSD for a few years.= =0AI would like to know if there is any benchmark or real world performance= experience about this setting.=0AFor Linux, I have heard people having iss= ues with the THP (transparent huge page) on Linux.=0A=0A2.=0AI have seen so= mewhere that superpages support was being developed in HEAD too.=0AAny insi= ght on it?=0A=0ARegards,=0APatrick=0A From owner-freebsd-hackers@FreeBSD.ORG Thu Sep 19 02:44:38 2013 Return-Path: <owner-freebsd-hackers@FreeBSD.ORG> Delivered-To: freebsd-hackers@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTP id B0750D90 for <freebsd-hackers@freebsd.org>; Thu, 19 Sep 2013 02:44:38 +0000 (UTC) (envelope-from nonesuch@longcount.org) Received: from mail-we0-f175.google.com (mail-we0-f175.google.com [74.125.82.175]) (using TLSv1 with cipher ECDHE-RSA-RC4-SHA (128/128 bits)) (No client certificate requested) by mx1.freebsd.org (Postfix) with ESMTPS id 4C0E32A44 for <freebsd-hackers@freebsd.org>; Thu, 19 Sep 2013 02:44:38 +0000 (UTC) Received: by mail-we0-f175.google.com with SMTP id q59so7432650wes.34 for <freebsd-hackers@freebsd.org>; Wed, 18 Sep 2013 19:44:29 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:content-type; bh=CAySSB//BpsTdRBIXQlnGSQAGpbV6UhtNvf2wMbQZrU=; b=NFI+dmQaVrNwljk6jVzsx+SPbLnUxJ7umcHATyF2+aaxGFr+T4oiHJn0GhfYvi/YbL ec6XnA6ac2U+6x7LrXNaugegqYuWJUcLfhjtvOsURiPoR5PVlEWssM4yvs73Vs5OtOPU n8xprCVkvEX14bJkNpEiupaeyO27ZjmklRoIugeKArjmwYkAiq0o+GpFFgKuRNhcTZfa 8T+nDE+g6BzgTGxbMcUPwPLSE+mlbG3ShzXu9A75ZvE+Ms2D58i+tTYNrXdf+61+3oib /p0J09QZAPjsugGxwCW3UF1nCUQI5haT833DSWGB3vtbQSXT781xDku4c2ESKimJFNPu BnnA== X-Gm-Message-State: ALoCoQnO9kK5qYKfcCzSjAJ2s2d7vjoI8UAGk54Lr5640lv5I8rzn5Co1k2q1rhOAJYoHoMqmS8i MIME-Version: 1.0 X-Received: by 10.180.185.166 with SMTP id fd6mr9733766wic.5.1379558312469; Wed, 18 Sep 2013 19:38:32 -0700 (PDT) Received: by 10.217.143.135 with HTTP; Wed, 18 Sep 2013 19:38:32 -0700 (PDT) X-Originating-IP: [74.88.191.50] In-Reply-To: <20130917181101.GB97298@dan.emsphone.com> References: <CAMXt9NYZNRqzuCmY4dgQNFaGGO5ruB2o-gBgf0Vhr_dnwjb5NQ@mail.gmail.com> <20130917181101.GB97298@dan.emsphone.com> Date: Wed, 18 Sep 2013 22:38:32 -0400 Message-ID: <CAMXt9NZ+rR17CaRer6YjDPXYVQ3TCKSJ66=11c4TGbox9Kh+aA@mail.gmail.com> Subject: Re: FreeBSD SNMP OID Question From: Mark Saad <nonesuch@longcount.org> To: "freebsd-hackers@freebsd.org" <freebsd-hackers@freebsd.org> Content-Type: text/plain; charset=UTF-8 X-Content-Filtered-By: Mailman/MimeDel 2.1.14 X-BeenThere: freebsd-hackers@freebsd.org X-Mailman-Version: 2.1.14 Precedence: list List-Id: Technical Discussions relating to FreeBSD <freebsd-hackers.freebsd.org> List-Unsubscribe: <http://lists.freebsd.org/mailman/options/freebsd-hackers>, <mailto:freebsd-hackers-request@freebsd.org?subject=unsubscribe> List-Archive: <http://lists.freebsd.org/pipermail/freebsd-hackers>; List-Post: <mailto:freebsd-hackers@freebsd.org> List-Help: <mailto:freebsd-hackers-request@freebsd.org?subject=help> List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-hackers>, <mailto:freebsd-hackers-request@freebsd.org?subject=subscribe> X-List-Received-Date: Thu, 19 Sep 2013 02:44:38 -0000 On Tue, Sep 17, 2013 at 2:11 PM, Dan Nelson <dnelson@allantgroup.com> wrote: > In the last episode (Sep 17), Mark Saad said: > > Can someone shed some light on a OID mystery I have. I am using cacti > > to trend some snmp data off a bunch of FreeBSD servers. > > > > I noticed someone added a graph to a cluster for UCDavis - ssRawSwapIn / > > UCDavis - ssRawSwapOut . The OIDs are .1.3.6.1.4.1.2021.11.62 / > > .1.3.6.1.4.1.2021.11.63 Their description is Number of blocks swapped > > in / Number of blocks swapped out . > > > # snmpwalk -c MyPassword -v2c -Of server00 .1.3.6.1.4.1.2021.11.62.0 > > .iso.org.dod.internet.private. > > enterprises.ucdavis.systemStats.ssRawSwapIn.0 > > = Counter32: 3588 > > That's a counter, so it's reporting the total number of pageins since boot > (or since snmp started, depending on the particular value you're fetching). > Cacti should be able to poll that OID and graph the difference over time to > show pageins/sec. > > > -- > Dan Nelson > dnelson@allantgroup.com > Dan I guess to better refine the question , what is raw swap vs the sysctl vm.stats.vm.v_swappgs{out/in} . I see that net-snmpd has ssSwapOut and ssRawSwapOut . where raw is the current value and "cooked" (ssswap{out/in}) is the average value . I am just at a loss when I am trying to debug this graph issue as the "cooked" oid returns negative ints and the raw returns positive ints, but the sysctrls and top show no usage ? Has anyone seen this before ? -- mark saad | nonesuch@longcount.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1379520488.49964.YahooMailNeo>