Date: Tue, 15 Feb 2005 17:27:51 -0500 From: Bosko Milekic <bosko.milekic@gmail.com> To: Max Laier <max@love2party.net> Cc: Borja Marcos <borjamar@sarenet.es> Subject: Re: MBUF statistics Message-ID: <bbebbd3d0502151427b11c3d3@mail.gmail.com> In-Reply-To: <200502151655.09800.max@love2party.net> References: <02266ed294653cda2452e7ac0a713c2c@sarenet.es> <200502151655.09800.max@love2party.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 15 Feb 2005 16:54:52 +0100, Max Laier <max@love2party.net> wrote: > On Tuesday 15 February 2005 12:38, Borja Marcos wrote: > > Hello, > > > > Looking at the mbuf statistics available in FreeBSD 4 and FreeBSD 5 I > > can see that the statistics available in FreeBSD 5 are, surprisingly, > > much less comprehensive. Is there any other place where I can find out > > how many mbuf requests have been done, how many of them have waited, > > how many have failed, etc? > > I use "$vmstat -z | grep Mbuf". The netstat -m output is broken, because > fixing this would impose an additional atomic operation on each alloc/free > which is a real performance killer. Yeah, unfortunately statistics are too hard to do completely correctly right now (too hard on performance, that is). To make things worse, the more involved UMA zone design used for Mbuf and Cluster allocations means that some of the UMA zone statistics you get with "vmstat -z" are not entirely accurate. The UMA zone statistics code needs to be changed to accomodate this structure, but in addition to that a way to make statistics gathering cheaper would be nice, because currently doing a 'vmstat -z' is really terrible for performance. Which reminds me... those of you doing benchmarks, please don't run 'vmstat -z' while you're doing them, it might skew/pessimize your results. -- Bosko Milekic - If I were a number, I'd be irrational. Contact Info: http://bmilekic.unixdaemons.com/contact.txt
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bbebbd3d0502151427b11c3d3>