Manually Rejoining Independent Clusters

Learn the consideration and needed steps to rejoin independent clusters in your environment.

Before you manually rejoin independent clusters, you must decide which cluster has the database and other files you want to preserve. The active node in the cluster with the chosen database becomes the new active node in the single cluster.

Important: This procedure is necessary only if you have disabled the self-healing mechanism described in The Self-Healing Mechanism.

To manually rejoin independent clusters after a WAN link failure:

  1. Decide which of the two clusters has the database you want to use going forward.
    The cluster with the database and other files you want to keep is the preserved cluster. The other cluster is the abandoned cluster.
  2. Restore the WAN link.
    The two clusters continue running independently.
  3. Use the Vocera Control Panel to force a failover on the active node of the abandoned cluster. See Using the Cluster Menu.

    The following events occur:

    • The standby node of the abandoned cluster enters discovery mode, sees the active node of the preserved cluster, and comes online as a standby node for it. This new standby node then performs a remote restore from the active node.

    • The formerly active node of the abandoned cluster restarts, enters discovery mode, sees the active node of the preserved cluster, and comes online as a standby node for it. This new standby node then performs a remote restore from the active node.

    • The badges that were connected to the abandoned cluster find the active node and connect to it, because the active node is still in their cluster list.

    • For Vocera Telephony Gateway server or Vocera Report Server machines that were previously connected to an abandoned cluster, you must check the cluster list, find the active node, and connect to it.

    Note: Do not use the Force Restart button on the Cluster page of the System screen to restart the active node of the abandoned cluster. Use the Vocera Control Panel to force all services to restart.

A remote restore of a very large database (50,000 spoken names) across a WAN may take 20 minutes or more, depending on the actual size of the database, the speed of the WAN, the available bandwidth, and other issues outside the control of Vocera.

Note: You must have a high-speed WAN link that meets the latency requirements described in the Vocera Infrastructure Planning Guide to support a geographically distributed cluster.