From owner-freebsd-hackers@FreeBSD.ORG Wed Feb 2 11:10:47 2005 Return-Path: Delivered-To: freebsd-hackers@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id BEB8616A4CF for ; Wed, 2 Feb 2005 11:10:47 +0000 (GMT) Received: from sollube.sarenet.es (sollube.sarenet.es [192.148.167.16]) by mx1.FreeBSD.org (Postfix) with ESMTP id 162E343D5E for ; Wed, 2 Feb 2005 11:10:46 +0000 (GMT) (envelope-from borjamar@sarenet.es) Received: from [127.0.0.1] (borja.sarenet.es [192.148.167.77]) by sollube.sarenet.es (Postfix) with ESMTP id 5B8D7150D for ; Wed, 2 Feb 2005 12:10:42 +0100 (CET) Mime-Version: 1.0 (Apple Message framework v619.2) Content-Transfer-Encoding: 7bit Message-Id: Content-Type: text/plain; charset=US-ASCII; format=flowed To: freebsd-hackers@freebsd.org From: Borja Marcos Date: Wed, 2 Feb 2005 12:10:41 +0100 X-Mailer: Apple Mail (2.619.2) Subject: Devilator - performance monitoring for FreeBSD X-BeenThere: freebsd-hackers@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Technical Discussions relating to FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 02 Feb 2005 11:10:47 -0000 Hello, I'm writing a performance monitoring data collector for Orca (www.orcaware.com) for FreeBSD 4- and 5-. I'm not sure about the correct values in the process description to get a picture as accurate as possible of the cpu usage of different processes. I've seen that top uses p_runtime (FreeBSD 5 and FreeBSD 4), but I'm not sure if the value would be really useful. You can see a snapshot of the work in progress at: ftp://borja.sarenet.es/pub/freebsd4-devilator.pdf ftp://borja.sarenet.es/pub/freebsd5-devilator.pdf I'm intending to do something more complete than the classical "orcallator" for Solaris. Namely, I am going to plot: - System processes resource usage (hopefully useful to spot bottlenecks, and hopefully useful for the system developers) - Resource usage by a set of processes specified by the user. It will have a configuration file with {process name, regular expression} pairs. Processes whose name matches the regular expression will get their own graph with %user/%system, etc cpu times, and probably I/O statistics, memory statistics, so that you can know wether your (for example) smtpd processes are getting more resources, or the memory hogs are the httpd's, etc. - MBUF statistics - Network statistics (connections, TCP/UDP/ICMP statistics...) - Various caches and VM BTW, I'm having serious issues with a machine with very big directories, and I've been playing with the dirhash configuration, setting up a very big cache. It would be useful to have some statistics so that I can plot the number of hits/misses to that dirhash cache, etc. Please send me suggestions, ideas, problems seen in these examples. The software will obviously be released to the community, and I plan to make the first release available in one or two weeks. I know that there may be too many graphs in the page, and I will probably add some switches to turn graphs on/off. Best regards, Borja.