• How does VMware vSphere FDM (HA) select a master:

vSphere HA clusters uses an election process to determine which host is to be the master. This election process occurs any time the existing master fails, is shut down, or is placed into maintenance mode. It also occurs when vSphere HA is enabled, or when a management network partition occurs. The election process takes about 10 to 15 seconds.

The election process is defined by an algorithm with two published rules. For the first, the host with access to the greatest number of datastores wins. In the case of a tie, the second rule kicks in: The host with the lexically-highest Managed Object ID (MOID) is chosen. Care must be taken when attempting to rig this election because lexically here means, for example, that host-99 is in fact higher than host-100.

  • Is it possible to influence the HA/FDM master election:

This is not recommend or best practise, but there actually is a way to make it very likely a certain host becomes the HA master.

The option is:

fdm.nodeGoodness

How does it works:

When a master election is held, the FDMs exchange a
goodness value, and the FDM with the largest goodness value is elected master.
Ties are broken using the host IDs assigned by vCenter Server. This parameter
can be used to override the computed
goodness value for a given FDM. To force a specific host to be
elected master each time an election is held and the host is active, set this
option to a large positive value. This option should not be
specified at the cluster level.

You can set this by manually editing the following file on the host where you want to increase the chances of winning the election:

/etc/opt/vmware/fdm/fdm.cfg

More details about this can be found in this KB, which also lists many other advanced options.

Advertisement