2010/2/17 <darren.mans...@opengi.co.uk>: > On Tue, 2010-02-16 at 22:24 +0100, Michael Schwartzkopff wrote: > Yes. Perhaps your could add some lines to your ldirectord OCF script t > ocheck >> the state of the sync daemon. >> >> But this setup works when I have my classes. > > > > Personally I have ipvsadm init script chkconfig'd on so it will start at > boot time, then the following in /etc/ipvsadm.rules on node1: > > --start-daemon master --syncid 50 > --start-daemon backup --syncid 51 > > Then the following in /etc/ipvsadm.rules in node2: > > --start-daemon master --syncid 51 > --start-daemon backup --syncid 50 > > It could be easily added to the ldirectord RA and I keep meaning to do > it (like lots of other things). > > Aside from this I have the cluster start ldirectord and an IPAddr2 RA > using the param lvs_support=true, add the VIP to lo on both nodes, set > the LVS ARP ignore sysctl params and you're away. > > This gives you load-balancing for free on top of HA, not sure why more > people don't do it TBH. > http://www.clusterlabs.org/wiki/Load_Balanced_MySQL_Replicated_Cluster#M > ySQL_Configuration > > Regards, > Darren
I already have it working like that, although I omitted the syncid parameter (seems to not being needed according to some mail posts, don't have the link at hand) and specified an alternative broadcast interface (the cluster interconnect). In Pacemaker I have a group that manages de floating IP (IPAddr2+lvs_support) and ldirectord. The real servers have the "LVS ARP ignore" magic activated and the real IP on the loopback. Regards, -- Ciro Iriarte http://cyruspy.wordpress.com -- _______________________________________________ Linux-HA mailing list Linux-HA@lists.linux-ha.org http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems