Return-Path: Delivered-To: apmail-httpd-test-dev-archive@httpd.apache.org Received: (qmail 66334 invoked by uid 500); 29 Jun 2003 12:31:19 -0000 Mailing-List: contact test-dev-help@httpd.apache.org; run by ezmlm Precedence: bulk Reply-To: test-dev@httpd.apache.org list-help: list-unsubscribe: list-post: Delivered-To: mailing list test-dev@httpd.apache.org Received: (qmail 66314 invoked from network); 29 Jun 2003 12:31:18 -0000 Date: Sun, 29 Jun 2003 13:56:12 +0200 From: Jacek Prucia To: test-dev@httpd.apache.org Subject: Re: flood - Can it go faster? Message-Id: <20030629135612.7675ebf7.jacek.prucia@acn.waw.pl> In-Reply-To: References: X-Mailer: Sylpheed version 0.9.0 (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Spam-Rating: daedalus.apache.org 1.6.2 0/1000/N On Tue, 24 Jun 2003 19:35:49 +0000 (GMT) Jeremy Brown wrote: > > Hi, Hi, > I have been playing with flood for a while. And have been looking at > writing up some additional documentation. I'll be commiting stuff at http://icarus.apache.org/~jacekp/flood-docs.html to CVS repo shortly, so please hold on just a day, or two. [..] > Are there any tips to get more throughput on flood? Please try figuring out what resource of your operating system seems to be exhausted. If memory, CPU time ussage are below maximum usage, and your system hasn't run out of socket descriptors -- then we have problem with flood that we need to investigate. Can you run gprof on your instance of flood and post stats here? That would help a lot... [...] > Also is there a way to turn off logging completely? Yep. Just remove element from and you'll get no output at all. Basically removing element most of the time causes flood to do nothing, as this is "generic" behaviour. > The new server will > have logging enable to catch all the information. Not really -- if your server (assuming Apache) segfaults under high load -- how would you tell what flood instance|farm|farmer|url caused that? > I am using shell scripts of flood since megaconglomerate is not operating > at this time from what I have tested. Are there any recommendations for > doing this more effectively? Or is megaconglomerate really working? Megaconglomerate is not working right now, but we'll get to that. You can start writting megaconglomerate support if you want (look at CONFIG file to see examples of how this could look like). > ps Here is the config file for flood that I have been using. You might want to switch to keepalive in your . Also please note that this setup won't generate all the load, as farmer is limited to 180 seconds time (