Table of ContentsView in Frames

Best practices for cf options

You should generally use the default options, since they represent the best choice for most configurations.

The following table lists best practices for the cf options available in Data ONTAP. These settings are changed with the options command. You use the man options command to display the man page with detailed descriptions of the options.

Option Recommended value and notes
cf.giveback.auto.cancel.on_network_failure ON

Leave this option on to avoid automatic giveback without assurance that the partner's network interfaces are fully operational.

cf.giveback.auto.cifs.terminate.minutes 5 minutes (default value)
cf.giveback.auto.delay.seconds 600 seconds (default value)
cf.giveback.auto.enable OFF (default value)

Leave this option off so that in cases other than takeover due to reboot or panic, you can verify that the failure that caused the takeover is fixed before performing a giveback.

Each node in an HA pair can have a different setting for this option

cf.giveback.auto.after.panic.takeover ON (default value)
cf.giveback.auto.override.vetoes OFF (default value)

Each node in an HA pair can have a different setting for this option.

cf.giveback.check.partner

ON (default value)

Leave on to reduce downtime caused by a unsuccessful giveback.

Each node in an HA pair can have a different setting for this option.

cf.hw_assist.enable ON (default value)

The node must support remote management through a Remote LAN Management card or Service Processor to enable the hardware-assisted takeover.

cf.hw_assist.partner.address When using hardware-assisted takeover, the value of this option should be equal to partner's node management IP address.
cf.hw_assist.partner.port 4444 (default value)
cf.takeover.change_fsid ON (default value)

Each node in an HA pair can have a different setting for this option.

cf.takeover.detection.seconds 15 seconds (default value)

If sk.process.timeout.override has been manually set, it is strongly advised that this option is set to a value larger than or equal to the value of sk.process.timeout.override+5.

cf.takeover.on_failure ON (default value)

Changing the value on one node automatically changes the value on the partner.

cf.takeover.on_network_interface_failure OFF (default value)

Changing the value on one node automatically changes the value on the partner.

cf.takeover.on_network_interface_failure.policy Use all_nics to avoid spurious takeovers due to any one network interface failure.
cf.takeover.on_panic ON (default value)

Use caution when manually changing the option value. In normal circumstances, leave this option on to avoid outages due to panics.

Changing the value on one node automatically changes the value on the partner.

cf.takeover.on_reboot For systems not licensed for FC or iSCSI, the takeover on reboot outage and subsequent giveback outage are shorter than the longer outage incurred by the reboot. You should use the default value of ON for these systems.

For systems licensed for FC or iSCSI, you should use the default value of OFF. In these systems, the reboot outage is shorter than the combined takeover and giveback outage.

Changing the value on one node automatically changes the value on the partner.

cf.takeover.on_short_uptime ON (default value)

Leave this option on to avoid outages due to failures occurring early in the node's boot process.

Changing the value on one node automatically changes the value on the partner.

cf.takeover.use_mcrc_file OFF (default value)

Use only in MetroCluster environments.