On 02/25/2011 04:01 AM, Raymond Fung wrote: > Dear all, > > For a two nodes cluster configuration, in the previous days with the > "heartbeat" tool, I can configure additional communication channels over > extra LAN and serial ports interfaces for communication redundancy. > Also, I shall configure ping nodes or ping groups for the cluster nodes > to compare their "capability" of communication with the outside world to > determine which node shall act as the primary node for a active-passive > fail-over configuration. How can I program these or the similar with > Corosync and pacemaker ? > > With corosync for a two nodes cluster, the default simple configuration > with multicast over Ethernet interface cannot tell which node has been > down when the communication between the two nodes broke down. This could > be the LAN interface on one of the nodes (or the LAN cable or port on > the LAN switch), or some route settings problem in between. This shall > create a split-brain situation and each node shall announce itself as > the only available working node and take over the sharing resources at > the same time. > > Anyone can point me to the right direction ? > > Regards, > Raymond. >
Kernel bonding driver run in mode=1 seems appropriate for this use case. Regards -steve > _______________________________________________ > Openais mailing list > [email protected] > https://lists.linux-foundation.org/mailman/listinfo/openais _______________________________________________ Openais mailing list [email protected] https://lists.linux-foundation.org/mailman/listinfo/openais
