Product Documentation

Forcing a Node to Fail Over

Sep 01, 2016

You might want to force a failover if, for example, you need to replace or upgrade the primary node. You can force failover from either the primary or the secondary node. A forced failover is not propagated or synchronized. To view the synchronization status after a forced failover, you can view the status of the node.

A forced failover fails in any of the following circumstances:
  • You force failover on a standalone system.
  • The secondary node is disabled.
  • The secondary node is configured to remain secondary.

The NetScaler appliance displays a warning message if it detects a potential issue when you run the force failover command. The message includes the information that triggered the warning, and requests confirmation before proceeding.

You can force a failover on a primary node, secondary node, and when nodes are in listen mode.

  • Forcing Failover on the Primary Node.

    If you force failover on the primary node, the primary becomes the secondary and the secondary becomes the primary. Forced failover is possible only when the primary node can determine that the secondary node is UP.

    If the secondary node is DOWN, the force failover command returns the following error message: "Operation not possible due to invalid peer state. Rectify and retry."

    If the secondary system is in the claiming state or inactive, it returns the following error message: "Operation not possible now. Please wait for system to stabilize before retrying."

  • Forcing Failover on the Secondary Node.

    If you run the force failover command from the secondary node, the secondary node becomes primary and the primary node becomes secondary. A force failover can occur only if the secondary node’s health is good and it is not configured to stay secondary.

    If the secondary node cannot become the primary node, or if secondary node was configured to stay secondary (using the STAYSECONDARY option), the node displays the following error message: "Operation not possible as my state is invalid. View the node for more information."

  • Forcing Failover When Nodes Are in Listen Mode.

    When the two nodes of an HA pair are running different versions of the system software, the node running the higher version switches to the listen mode. In this mode, neither command propagation nor synchronization works.

    Before upgrading the system software on both nodes, you should test the new version on one of the nodes. To do this, you need to force a failover on the system that has already been upgraded. The upgraded system then takes over as the primary node, but neither command propagation or synchronization occurs. Also, all connections need to be re-established.

To force failover on a node by using the command line interface

At the command prompt, type:

force HA failover

To force failover on a node by using the configuration utility

Navigate to System > High Availability and, on the Nodes tab, select the node, in the Action list, select Force Failover.