From owner-freebsd-cluster@FreeBSD.ORG Thu Mar 27 16:27:49 2008 Return-Path: Delivered-To: freebsd-cluster@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 0F9EF1065676 for ; Thu, 27 Mar 2008 16:27:49 +0000 (UTC) (envelope-from nbiondi@iridiumir.com) Received: from an-out-0708.google.com (an-out-0708.google.com [209.85.132.248]) by mx1.freebsd.org (Postfix) with ESMTP id CA9F38FC2B for ; Thu, 27 Mar 2008 16:27:48 +0000 (UTC) (envelope-from nbiondi@iridiumir.com) Received: by an-out-0708.google.com with SMTP id c14so146680anc.13 for ; Thu, 27 Mar 2008 09:27:48 -0700 (PDT) Received: by 10.100.133.11 with SMTP id g11mr3708212and.100.1206635268070; Thu, 27 Mar 2008 09:27:48 -0700 (PDT) Received: by 10.100.127.19 with HTTP; Thu, 27 Mar 2008 09:27:48 -0700 (PDT) Message-ID: Date: Thu, 27 Mar 2008 09:27:48 -0700 From: "Nathan Biondi" To: elliott@c7.ca In-Reply-To: <1206571714.3820.129.camel@kensho.c7.ca> MIME-Version: 1.0 References: <1206571714.3820.129.camel@kensho.c7.ca> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Content-Filtered-By: Mailman/MimeDel 2.1.5 Cc: freebsd-cluster@freebsd.org Subject: Re: haproxy and heartbeat X-BeenThere: freebsd-cluster@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Clustering FreeBSD List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 27 Mar 2008 16:27:49 -0000 On Wed, Mar 26, 2008 at 3:48 PM, Elliott Perrin wrote: > Hi Nathan, > > I have a FreeBSD cluster running with 4 apache web servers, 2 DB > servers, 2 NFS servers and 2 firewalls out in front All machines run > FreeBSD. I built this using CARP as the basis of communications for > redundancy / load balancing with a unique configuration that has been > running for a couple of years now. This cluster is fully redundant with > automatic failovers built in in case of service / host outages. I am > not using HAProxy or heartbeat, but I would be able to help you (I have > used those packages before) if you can provide some more information > about what you are trying to do, current configurations and such. If > you would like to communicate privately regarding this feel free to > drop me an email. > > As far as changing that sysctl variable, putting that line into > sysctl.conf will cause that value to be set at boot time. If you want > to change the current value of that MIB just do > > sysctl net.ipv4.ip_nonlocal_bind=1 > > from the command line as root, however I do not see that MIB on any of > my 7-STABLE or 6-STABLE machines. Where did it say that you should make > that sysctl adjustment, and what version of FreeBSD are you running? > > Cheers, > Elliott Perrin > elliott@c7.ca I currently have 2 DB servers running as data nodes for my mysql cluster, 2 web servers running apache with SSL and 2 servers running as primary and secondary management nodes as well as loadbalancers running haproxy and heartbeat that I am finishing the configuration currently. I want the data nodes and web servers loadbalanced with auto-failover. All machines are running 6.2. I was using this Debian tutorial: http://www.howtoforge.com/high-availability-load-balancer-haproxy-heartbeat-debian-etchas a reference for my configuration and it is here that it says to add that sysctl line to sysctl.conf and the run sysctl -p which appears to not exist in FreeBSD. When I run the sysctl net.ipv4.ip_nonlocal_bind=1from command line it tells me that net.ipv4... does not exist. I did add this line to sysctl.conf and restarted but it gives me an error at boot time that the specific oid does not exist. This leads me to believe there are other differences in sysctl between FreeBSD and linux distributions, but have not been able to discover what those are at this time. I had been thinking about using CARP but it seemed like I would need to configure pf and as I am still learning pf I am not fully confident in my ability to configure it properly. Is CARP the best solution for this? Thank you again, Nathan Biondi