Reasons that automatic takeover occurs

You can set options to control whether automatic takeovers occur due to different system errors. In some cases, automatic takeover occurs by default unless you disable the option, and in some cases automatic takeover cannot be prevented.

Takeovers can happen for several reasons. Some system errors cause a takeover; for example, when a system in an HA pair loses power, it automatically fails over to the other node.

However, for some system errors, a takeover is optional, depending on how you set up your HA pair. The following table outlines which system errors can cause a takeover to occur, and whether you can configure the HA pair for that error.

System error Option used to configure Default value Notes
A node undergoes a system failure and cannot reboot. cf.takeover.on_failure set to on On You should leave this option enabled unless instructed otherwise by technical support.
A node undergoes a software or system failure leading to a panic. cf.takeover.on_panic set to on On If the cf.takeover.on_panic option is set to on, a node panic will cause an automatic takeover.

If the cf.takeover.on_panic option is set to off, a node panic will not cause an automatic takeover. You should not turn this option off unless you are instructed by technical support to do so.

A node reboots. cf.takeover.on_reboot set to on On (unless FC or iSCSI is licensed)
Note: In releases prior to Data ONTAP 8.0, the system would only take over the partner after reboot if the partner took longer than 90 seconds to boot.
If a takeover is performed because of the partner node rebooting, then an automatic giveback will be performed, regardless of the setting of the cf.giveback.auto.enable option.
All the network interface cards (NICs) or interface groups enabled for negotiated failover on a node failed. cf.takeover.on_network_ interface_failure set to on, cf.takeover.on_network_ interface_failure.policy set to all_nics Takeover on network failure disabled (by default) To enable a network interface for negotiated failover, you use the ifconfig if_name nfo command. For more information, see the Data ONTAP MultiStore Management Guide for 7-Mode.
One or more of the NICs or interface groups enabled for negotiated failover failed.
Note: If interfaces fail on both nodes in the HA pair, takeover won't occur.
cf.takeover.on_network_ interface_failure set to on

cf.takeover.on_network_ interface_failure.policy set to any_nic

Takeover on network failure disabled (by default) To enable a network interface or interface group for negotiated failover, you use the ifconfig if_name nfo command. For more information, see the Data ONTAP MultiStore Management Guide for 7-Mode.
A node fails within 60 seconds of booting up. cf.takeover.on_short_uptime set to on On Changing the value of this option on one node automatically updates the option on the partner node.
A node cannot send heartbeat messages to its partner. Not applicable Not applicable You cannot prevent this condition from causing a takeover.
You halt one of the nodes without using the -f flag. Not applicable Not applicable You cannot prevent this condition from causing a takeover. If you include the -f flag, the takeover is prevented.
You initiate a takeover manually using the cf takeover command. Not applicable Not applicable You cannot prevent this condition from causing a takeover.