Install and maintain

Contributors dougthomp
PDFs

Shutdown the controllers - AFF A900

This procedure is for 2-node, non-MetroCluster configurations only. If you have a system with more than two nodes, see How to perform a graceful shutdown and power up of one HA pair in a 4-node cluster.

Before you begin

You need:

  • Local administrator credentials for ONTAP.

  • NetApp onboard key management (OKM) cluster-wide passphrase if using storage encryption.

  • SP/BMC accessability for each controller.

  • Stop all clients/host from accessing data on the NetApp system.

  • Suspend external backup jobs.

  • Necessary tools and equipment for the replacement.

Note If the system is a NetApp StorageGRID or ONTAP S3 used as FabricPool cloud tier, refer to the Gracefully shutdown and power up your storage system Resolution Guide after performing this procedure.
Note If using FlexArray array LUNs, follow the specific vendor storage array documentation for the shutdown procedure to perform for those systems after performing this procedure.
Note If using SSDs, refer to SU490: (Impact: Critical) SSD Best Practices: Avoid risk of drive failure and data loss if powered off for more than two months

As a best practice before shutdown, you should:

Steps

  1. Log into the cluster through SSH or log in from any node in the cluster using a local console cable and a laptop/console.

  2. Turn off AutoSupport and indicate how long you expect the system to be off line:

    system node autosupport invoke -node * -type all -messages "MAINT=8h Power Maintenance"

  3. Identify the SP/BMC address of all nodes:

    system service-processor show -node * -fields address

  4. Exit the cluster shell: exit

  5. Log into SP/BMC over SSH using the IP address of any of the nodes listed in the output from the previous step.

    If your’re using a console/laptop, log into the controller using the same cluster administrator credentials.

    Note Open an SSH session to every SP/BMC connection so that you can monitor progress.

  6. Halt all nodes in the cluster:

    system node halt -node * -skip-lif-migration-before-shutdown true -ignore-quorum-warnings true -inhibit-takeover true.

    Note For clusters using SnapMirror synchronous operating in StrictSync mode: system node halt -node * -skip-lif-migration-before-shutdown true -ignore-quorum-warnings true -inhibit-takeover true -ignore-strict-sync-warnings true

  7. Enter y for each controller in the cluster when you see Warning: Are you sure you want to halt node "cluster name-controller number"? {y|n}:

  8. Wait for each controller to halt and display the LOADER prompt.

  9. Turn off each PSU or unplug them if there is no PSU on/off switch.

  10. Unplug the power cord from each PSU.

  11. Verify that all controllers in the impaired chassis are powered down.