Hello, we are using hardware array and its vendor recommends the following setting in /etc/system: set sd:sd_max_throttle = <value> or set ssd:ssd_max_throttle = <value> Is it possible to monitor *somehow* whether the variable becomes sort of bottleneck ? Or how its value influences io traffic ? Regards przemol
przemolicc at poczta.fm wrote:> Hello, > > we are using hardware array and its vendor recommends the following setting in > /etc/system: > set sd:sd_max_throttle = <value> > or > set ssd:ssd_max_throttle = <value> >To find out a little more of what this feature does, have a read here: http://www.opensolaris.org/jive/thread.jspa?threadID=12523&tstart=0> > Is it possible to monitor *somehow* whether the variable becomes > sort of bottleneck ? Or how its value influences io traffic ? >You could probably monitor the performance using the standard tools, for example, iostat. What you are doing when you set these commands is effectivly limiting the total number of io requests to a particular device. This is "recommended" with arrays because they share out multiple "luns" over a single port and thus the number of io requests from the hosts may overwhelm the FEP.> Regards > przemol
Bruce, I know it is not elegant but I run iostat -xDn 30 and if I see wsvc_t time then I assume that sd_max_throttle is hurting me. rick On Tue, Nov 13, 2007 at 02:03:54PM +0000, Bruce McAlister wrote:> Date: Tue, 13 Nov 2007 14:03:54 +0000 > From: Bruce McAlister <bruce.mcalister at blueface.ie> > Subject: Re: [dtrace-discuss] sd_max_throttle > In-reply-to: <20071113115433.GD5667 at lap.pgf.com.pl> > To: dtrace-discuss at opensolaris.org > Errors-to: dtrace-discuss-bounces at opensolaris.org > X-Complaints-to: usenet at ger.gmane.org > Precedence: list > X-BeenThere: dtrace-discuss at opensolaris.org > Delivered-to: dtrace-discuss at opensolaris.org > X-PMX-Version: 5.2.0.264296 > X-Brightmail-Tracker: AAAAAA=> X-Spam-Checker-Version: SpamAssassin 3.2.3 (2007-08-08) on > oss-mail1.opensolaris.org > X-Original-To: dtrace-discuss at opensolaris.org > X-Injected-Via-Gmane: http://gmane.org/ > X-Gmane-NNTP-Posting-Host: 89.100.145.104 > X-Mailman-Version: 2.1.9 > List-Post: <mailto:dtrace-discuss at opensolaris.org> > List-Subscribe: <http://mail.opensolaris.org/mailman/listinfo/dtrace-discuss>, > <mailto:dtrace-discuss-request at opensolaris.org?subject=subscribe> > List-Unsubscribe: > <http://mail.opensolaris.org/mailman/listinfo/dtrace-discuss>, > <mailto:dtrace-discuss-request at opensolaris.org?subject=unsubscribe> > List-Archive: <http://mail.opensolaris.org/pipermail/dtrace-discuss> > List-Help: <mailto:dtrace-discuss-request at opensolaris.org?subject=help> > List-Id: DTrace General Discussion <dtrace-discuss.opensolaris.org> > User-Agent: Thunderbird 2.0.0.6 (Windows/20070728) > X-Spam-Status: No, score=2.6 required=5.0 tests=RCVD_NUMERIC_HELO autolearn=no > version=3.2.3 > X-Spam-Level: ** > > przemolicc at poczta.fm wrote: > > Hello, > > > > we are using hardware array and its vendor recommends the following setting in > > /etc/system: > > set sd:sd_max_throttle = <value> > > or > > set ssd:ssd_max_throttle = <value> > > > > To find out a little more of what this feature does, have a read here: > > http://www.opensolaris.org/jive/thread.jspa?threadID=12523&tstart=0 > > > > > Is it possible to monitor *somehow* whether the variable becomes > > sort of bottleneck ? Or how its value influences io traffic ? > > > > You could probably monitor the performance using the standard tools, for > example, iostat. > > What you are doing when you set these commands is effectivly limiting > the total number of io requests to a particular device. This is > "recommended" with arrays because they share out multiple "luns" over a > single port and thus the number of io requests from the hosts may > overwhelm the FEP. > > > Regards > > przemol > > _______________________________________________ > dtrace-discuss mailing list > dtrace-discuss at opensolaris.org-- Rickey C. Weisner Software Development and Performance Specialist Sun Microsystems, INC cell phone: 615-308-1147 email: rick.weisner at sun.com
On Tue, Nov 13, 2007 at 02:03:54PM +0000, Bruce McAlister wrote:> przemolicc at poczta.fm wrote: > > Hello, > > > > we are using hardware array and its vendor recommends the following setting in > > /etc/system: > > set sd:sd_max_throttle = <value> > > or > > set ssd:ssd_max_throttle = <value> > > > > To find out a little more of what this feature does, have a read here: > > http://www.opensolaris.org/jive/thread.jspa?threadID=12523&tstart=0Thank you for the url. There is even more at: http://forums.hds.com/index.php?showtopic=528&hl=sd_max_throttle I am looking for a way to monitor how it influences the whole io traffic. The same way as when using ufs_LW/ufs_HW. When number of writes exceed value of ufs_HW, the writes are deffered until only ufs_LW bytes are outstanding _and_ (what is more important !) the kernel variable ufs_throttles is increased. Maybe there is something similar in sd driver world ? Or to put it simply: can I monitor (s)sd queue (number of waiting requests in queue) using DTrace ? Regards przemol ---------------------------------------------------------------------- "Kup bilet na najlepsze zawody Freestyle Motocross - DIVERSE Night of the Jupms!" http://link.interia.pl/f1c5f
On Wed, Nov 14, 2007 at 08:59:27AM +0100, przemolicc at poczta.fm wrote:> On Tue, Nov 13, 2007 at 02:03:54PM +0000, Bruce McAlister wrote: > > przemolicc at poczta.fm wrote: > > > Hello, > > > > > > we are using hardware array and its vendor recommends the following setting in > > > /etc/system: > > > set sd:sd_max_throttle = <value> > > > or > > > set ssd:ssd_max_throttle = <value> > > > > > > > To find out a little more of what this feature does, have a read here: > > > > http://www.opensolaris.org/jive/thread.jspa?threadID=12523&tstart=0 > > Thank you for the url. There is even more at: > http://forums.hds.com/index.php?showtopic=528&hl=sd_max_throttle > > I am looking for a way to monitor how it influences the whole io traffic. > The same way as when using ufs_LW/ufs_HW. When number of writes exceed > value of ufs_HW, the writes are deffered until only ufs_LW bytes are > outstanding _and_ (what is more important !) the kernel variable > ufs_throttles is increased. Maybe there is something similar in sd > driver world ? > > Or to put it simply: can I monitor (s)sd queue (number of waiting > requests in queue) using DTrace ?Or to make it even simplier: I would like to monitor all (or most ?) events described at http://wikis.sun.com/display/StorageDev/The+Solaris+OS+Queue+Throttle Or at least to monitor how IO requests are getting into (sd) queue for specyfic LUN, how long is queue at the moment, when IO is removed from queue, etc. I belive it is possible (what is impossible for DTrace ? ;-)) ) but ... Regards przemol -- http://przemol.blogspot.com/ ---------------------------------------------------------------------- Zobacz ocenzurowany wywiad z Jaroslawem Kaczynskim! Kliknij >>> http://link.interia.pl/f1c4c