Skip to main content

Prepare the nodes for upgrade

Contributors netapp-pcarriga

The controller replacement process begins with a series of prechecks. You also gather information about the original nodes for use later in the procedure and, if required, determine the type of self-encrypting drives that are in use.

Steps
  1. Begin the controller replacement process by entering the following command in the ONTAP command line:

    system controller replace start -nodes <node_names>

    Note You can only execute the system controller replace start command at the advanced privilege level: set -privilege advanced

    You will see output similar to the following example. The output displays the ONTAP version running on your cluster:

    Warning: 1. Current ONTAP version is 9.15.1
    
    2. Verify that NVMEM or NVRAM batteries of the new nodes are charged, and charge them if they are not. You need to physically check the new nodes to see if the NVMEM or NVRAM batteries are charged. You can check the battery status either by connecting to a serial console or using SSH, logging into the Service Processor (SP) or Baseboard Management Controller (BMC) for your system, and use the system sensors to see if the battery has a sufficient charge.
    
    Attention: Do not try to clear the NVRAM contents. If there is a need to clear the contents of NVRAM, contact NetApp technical support.
    
    3. If a controller was previously part of a different cluster, run wipeconfig before using it as the replacement controller.
    
    4. Note: This is not a MetroCluster configuration. Controller replacement supports only ARL based procedure.
    Do you want to continue? {y|n}: y
  2. Press y, you will see the following output:

    Controller replacement operation: Prechecks in progress.
    Controller replacement operation has been paused for user intervention.

    The system runs the following prechecks; record the output of each precheck for use later in the procedure:

    Precheck Description

    Cluster Health Check

    Checks all the nodes in the cluster to confirm they are healthy.

    Aggregate Relocation Status Check

    Checks whether an aggregate relocation is already in progress.
    If another aggregate relocation is in progress, the check fails.

    Model Name Check

    Checks whether the controller models are supported for this procedure.
    If the models are not supported, the task fails.

    Cluster Quorum Check

    Checks that the nodes being replaced are in quorum. If the nodes are not in quorum, the task fails.

    Image Version Check

    Checks that the nodes being replaced run the same version of ONTAP.
    If the ONTAP image versions are different, the task fails.
    The new nodes must have the same version of ONTAP 9.x installed on them that is installed on the original nodes. If the new nodes have a different version of ONTAP installed, you need to netboot the new controllers after you install them. For instructions on how to upgrade ONTAP, refer to References to link to Upgrade ONTAP.

    HA Status Check

    Checks if both the nodes being replaced are in a high- availability (HA) pair configuration.
    If storage failover is not enabled for the controllers, the task fails.

    Aggregate Status Check

    If the nodes being replaced own aggregates for which they are not the home owner, the task fails.
    The nodes should not own any non-local aggregates.

    Disk Status Check

    If any nodes being replaced have missing or failed disks, the task fails.
    If any disks are missing, refer to References to link to Disk and aggregate management with the CLI, Logical storage management with the CLI, and High Availability management to configure storage for the HA pair.

    Data LIF Status Check

    Checks if any of the nodes being replaced have non- local data LIFs.
    The nodes should not contain any data LIFs for which they are not the home owner. If one of the nodes contains non-local data LIFs, the task fails.

    Cluster LIF Status

    Checks whether the cluster LIFs are up for both nodes. If the cluster LIFs are down, the task fails.

    ASUP Status Check

    If ASUP notifications are not configured, the task fails.
    You must enable ASUP before beginning the controller replacement procedure.

    CPU Utilization Check

    Checks if the CPU utilization is more than 50% for any of the nodes being replaced.
    If the CPU usage is more than 50% for a considerable period of time, the task fails.

    Aggregate Reconstruction Check

    Checks if reconstruction is occurring on any data aggregates.
    If aggregate reconstruction is in progress, the task fails.

    Node Affinity Job Check

    Checks if any node affinity jobs are running.
    If node affinity jobs are running, the check fails.

  3. After the controller replacement operation is started and the prechecks are completed, the operation pauses enabling you to collect output information that you might need later when configuring node3.

    Note

    Before you start the upgrade, you migrate and re-home the cluster LIFs to two cluster ports per node if you have a system, such as an AFF 700, with the following configuration:

    • More than two cluster ports per node

    • A cluster interconnect card in slot4 in breakout mode to create ports e4a, e4b, e4c, and e4d, and ports e4e, e4f, e4g, and e4h

    Performing a controller upgrade with more than two cluster ports per node might result in missing cluster LIFs on the new controller after the upgrade.

    For more information, see the Knowledge Base article How to delete unwanted or unnecessary cluster LIFs.

  4. Run the below set of commands as directed by the controller replacement procedure on the system console.

    From the serial port connected to each node, run and save the output of the following commands individually:

    • vserver services name-service dns show

    • network interface show -curr-node <local> -role <cluster,intercluster,node-mgmt,cluster-mgmt,data>

    • network port show -node <local> -type physical

    • service-processor show -node <local> -instance

    • network fcp adapter show -node <local>

    • network port ifgrp show -node <local>

    • system node show -instance -node <local>

    • run -node <local> sysconfig

    • storage aggregate show -r

    • storage aggregate show -node <local>

    • volume show -node <local>

    • system license show -owner <local>

    • storage encryption disk show

    • security key-manager onboard show-backup

    • security key-manager external show

    • security key-manager external show-status

    • network port reachability show -detail -node <local>

    Note If NetApp Volume Encryption (NVE) or NetApp Aggregate Encryption (NAE) using the Onboard Key Manager (OKM) is in use, keep the key manager passphrase ready to complete the key manager resync later in the procedure.
  5. If your system uses self-encrypting drives, see the Knowledge Base article How to tell if a drive is FIPS certified to determine the type of self-encrypting drives that are in use on the HA pair that you are upgrading. ONTAP software supports two types of self-encrypting drives:

    • FIPS-certified NetApp Storage Encryption (NSE) SAS or NVMe drives

    • Non-FIPS self-encrypting NVMe drives (SED)

Correct aggregate ownership if an ARL precheck fails

If the Aggregate Status Check fails, you must return aggregates owned by the partner node to the home owner node and initiate the precheck process again.

Steps
  1. Return the aggregates currently owned by the partner node to the home owner node:

    storage aggregate relocation start -node source_node -destination destination-node -aggregate-list *

  2. Verify that neither node1 nor node2 still owns aggregates for which it is the current owner (but not the home owner):

    storage aggregate show -nodes node_name -is-home false -fields owner-name, home-name, state

    The following example shows the output of the command when a node is both the current owner and home owner of aggregates:

    cluster::> storage aggregate show -nodes node1 -is-home true -fields owner-name,home-name,state
    aggregate   home-name  owner-name  state
    ---------   ---------  ----------  ------
    aggr1       node1      node1       online
    aggr2       node1      node1       online
    aggr3       node1      node1       online
    aggr4       node1      node1       online
    
    4 entries were displayed.

After you finish

You must restart the controller replacement process:

system controller replace start -nodes node_names

License

For detailed information about ONTAP licensing, refer to License management.

Note Using unlicensed features on the controller might put you out of compliance with your license agreement.