• If you have a cluster with more than two controllers, it must be in quorum. If the cluster is not in quorum or
a healthy controller shows
false
for eligibility and health, you must correct the issue before shutting down
the impaired controller; see the
Administration overview with the CLI
.
• If AutoSupport is enabled, suppress automatic case creation by invoking an AutoSupport message:
system node autosupport invoke -node * -type all -message
MAINT=number_of_hours_downh
The following AutoSupport message suppresses automatic case creation for two hours:
cluster1:*>
system node autosupport invoke -node * -type all -message MAINT=2h
Steps
1. If your system has two controller modules, disable the HA pair.
If your system is running
clustered ONTAP with…
Then…
Two controllers in the cluster
cluster ha modify -configured false
storage failover
modify -node node0 -enabled false
More than two controllers in the
cluster
storage failover modify -node node0 -enabled false
2. Halt the controller, pressing
y
when you are prompted to confirm the halt:
system node halt -node
node_name
The confirmation message looks like the following:
Warning: This operation will cause controller "node-name" to be marked
as unhealthy. Unhealthy nodes do not participate in quorum voting. If
the controller goes out of service and one more controller goes out of
service there will be a data serving failure for the entire cluster.
This will cause a client disruption. Use "cluster show" to verify
cluster state. If possible bring other nodes online to improve the
resiliency of this cluster.
Do you want to continue? {y|n}:
You must perform a clean system shutdown before replacing the chassis to avoid losing
unwritten data in the nonvolatile memory (NVMEM/NVRAM). Depending on your system, if
the NVMEM/NVRAM LED is flashing, there is content in the NVMEM/NVRAM that has not
been saved to disk. You need to reboot the controller and start from the beginning of this
procedure. If repeated attempts to cleanly shut down the controller fail, be aware that you
might lose any data that was not saved to disk.
3. Where applicable, halt the second controller to avoid a possible quorum error message in an HA pair
configuration:
system node halt -node
second_node_name
-ignore-quorum-warnings true
-skip-lif-migration-before-shutdown true
466