ipfail

ipfail is an add-on designed to interface with Linux-HA. It uses the Linux-HA API to communicate with the nodes[1] in the cluster[2]. The default setup of Linux-HA will not failover[3] resources[4] simply if an interface goes down. This is a reasonable scenario for many HA setups, however it is not acceptable in other instances, particularly when dealing with firewall pairs. If an interface goes down, or the network behind it is no longer reachable due to some other hardware failure, the resource holder must relinquish control to the backup, who will be able to take up the role as the primary. ipfail handles this by watching special ping nodes[5]. If a ping node is detected to have failed, ipfail will contact the other cluster member to see if it has better connectivity. If it does, then we can be reasonably sure that an interface/network connection has died and that we should failover our resources if we have any.

ipfail can now work with auto_failback[6], as of the 1.1.4 release of Heartbeat[7]. This allows a user to create active/active[8] clusters and still use ipfail.

Note that ipfail needs redundant communications media to work correctly - because it won't cause a failover on its own unless it can contact the other cluster member. In other words, if you're pinging on the same media as the only heartbeat channel configured, you're destined to be disappointed in ipfail.

Read how to configure ipfail[9].

For V2 style clusters you will have to look into pingd[10] instead.

ipfail is primarily the fault of KevinDwyer[11].

---

This information provided courtesy of the Linux-HA project at http://linux-ha.org/