• If you have a cluster with more than two nodes, it must be in quorum. If the cluster is not in quorum or a
healthy node shows
false
for eligibility and health, you must correct the issue before shutting down the
impaired node; 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 nodes in the cluster
cluster ha modify -configured false
storage failover
modify -node node0 -enabled false
More than two nodes in the
cluster
storage failover modify -node node0 -enabled false
2. Halt the node, pressing
y
when you are prompted to confirm the halt:
system node halt -node
node_name
The confirmation message looks like the following:
Warning: Rebooting or halting node
"node_name" in an HA-enabled cluster may result in client disruption or
data access
failure. To ensure continuity of service, use the "storage
failover takeover" command. Are you sure you want to halt node
"node_name"? {y|n}:
You must perform a clean system shutdown before replacing the chassis to avoid losing
unwritten data in the nonvolatile memory (NVMEM). If the NVMEM LED is flashing, there is
content in the NVMEM that has not been saved to disk. You need to reboot the node and
start from the beginning of this procedure. If repeated attempts to cleanly shut down the
node fail, be aware that you might lose any data that was not saved to disk.
3. Where applicable, halt the second node to avoid a possible quorum error message in an HA pair
configuration:
system node halt -node second_node_name -ignore-quorum-warnings true
Option 2: Controller is in a two-node MetroCluster configuration
To shut down the impaired node, you must determine the status of the node and, if
necessary, switch over the node so that the healthy node continues to serve data from
43
