Chapter 10. Cluster Quorum

A Red Hat Enterprise Linux High Availability Add-On cluster uses the votequorum service, in conjunction with fencing, to avoid split brain situations. A number of votes is assigned to each system in the cluster, and cluster operations are allowed to proceed only when a majority of votes is present. The service must be loaded into all nodes or none; if it is loaded into a subset of cluster nodes, the results will be unpredictable. For information on the configuration and operation of the votequorum service, see the votequorum(5) man page.

10.1. Configuring Quorum Options

There are some special features of quorum configuration that you can set when you create a cluster with the pcs cluster setup command. Table 10.1, “Quorum Options” summarizes these options.

Table 10.1. Quorum Options

When enabled, the cluster can suffer up to 50% of the nodes failing at the same time, in a deterministic fashion. The cluster partition, or the set of nodes that are still in contact with the nodeid configured in auto_tie_breaker_node (or lowest nodeid if not set), will remain quorate. The other nodes will be inquorate.
The auto_tie_breaker option is principally used for clusters with an even number of nodes, as it allows the cluster to continue operation with an even split. For more complex failures, such as multiple, uneven splits, it is recommended that you use a quorum device, as described in Section 10.5, “Quorum Devices”. The auto_tie_breaker option is incompatible with quorum devices.
When enabled, the cluster will be quorate for the first time only after all nodes have been visible at least once at the same time.
The wait_for_all option is primarily used for two-node clusters and for even-node clusters using the quorum device lms (last man standing) algorithm.
The wait_for_all option is automatically enabled when a cluster has two nodes, does not use a quorum device, and auto_tie_breaker is disabled. You can override this by explicitly setting wait_for_all to 0.
--last_man_standing When enabled, the cluster can dynamically recalculate expected_votes and quorum under specific circumstances. You must enable wait_for_all when you enable this option. The last_man_standing option is incompatible with quorum devices.
--last_man_standing_window The time, in milliseconds, to wait before recalculating expected_votes and quorum after a cluster loses nodes.
For further information about configuring and using these options, see the votequorum(5) man page.