Hi Andrew, On Wed, 2010-08-04 at 10:27 +0200, Andrew Beekhof wrote: > On Wed, Aug 4, 2010 at 12:56 AM, Brett Delle Grazie > <[email protected]> wrote: > > Hi, > > > > I have two nodes (RHEL 5.5) configured in a cluster > > (Corosync/Pacemaker). > > > > I have an IPaddr2 and Apache resources configured as clones on those > > systems (configuration is shown below). > > > > The IPaddr2 is configured for load balancing using ClusterIP with hash > > sourceip. > > > > My problem is that when failing back the ClusterIP after an initial > > failure test the cluster no longer receives traffic even though the > > ClusterIP appears to fail back correctly. > > > > Test as follows: > > > > 1. Test initial failure by adding location constraint to one node - this > > moves all ClusterIP elements to the opposing node: > > e.g. location loc_test cl_ipclust_0 -inf: node2 (will cause ClusterIP > > to move to node 1) > > > > 2. Observe in web browser traffic is still being sent to Apache and > > responses returned correctly > > > > 3. Now delete the test constraint and the ClusterIP will return back to > > node 2. (confirmed with ifconfig, cat /proc/net/ipt_CLUSTER/192.168.0.10 > > on both nodes - nodes respond with correct 'id', ping on both nodes to > > cluster IP resolves correctly) > > > > 4. Attempt to connect with web-browser to ClusterIP and this fails > > > > 5. Restart Apache clone and retry - still doesn't work > > > > The only way I can restore traffic is to stop the ClusterIP clone, wait > > a few seconds (5-10) and start it again. > > > > Anyone have any idea why this might be? > > An ARP issue perhaps?
And indeed that's exactly what it was. It turned out that the switches and the firewall needed static entries in the ARP tables for the cluster IP -> multicast mac address. Once they were put in place, clustering works perfectly. Thanks! > > > > > Any advice appreciated. > > > > Thanks, > > > > Configuration as follows: > > > > node node1 > > node node2 > > primitive apache_0 ocf:heartbeat:apache \ > > params configfile="/etc/httpd/conf/httpd.conf" > > httpd="/usr/sbin/httpd.worker" > > statusurl="http://localhost/server-status"; > > envfiles="/etc/sysconfig/httpd" \ > > op start interval="0" timeout="40" \ > > op stop interval="0" timeout="60" \ > > op monitor interval="10" timeout="40" > > primitive ipclust_0 ocf:intact:IPaddr2 \ > > params ip="192.168.0.10" nic="eth0" iflabel="1" > > clusterip_hash="sourceip" \ > > op monitor interval="10" timeout="5" > > clone cl_apache0 apache_0 \ > > meta globally_unique="false" interleave="true" > > clone cl_ipclust_0 ipclust_0 \ > > meta globally-unique="true" interleave="true" clone-node-max="2" > > clone-max="2" notify="true" \ > > params resource-stickiness="0" > > > > RHEL 5.5, kernel 2.6.18-194.8.1.el5 x86_64 > > Corosync 1.2.7-1.1.el5 > > Pacemaker 1.0.9.1-1.15.el5 > > resource-agents 1.0.3-2.6.el5 > > > > > > -- > > Best Regards, > > > > Brett Delle Grazie > > -- Best Regards, Brett Delle Grazie ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________ _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
