Pre-Merge Tasks

NOTE:
  • Array OS 4.X is required.
  • Merging an iSCSI array and a FC array together in the same group is not supported.
  • You cannot perform a pool merge between different array types, for example AFA, HFA and SFA.
    • To add an AFA to an existing HFA or SFA group and keep both (with their own pools), you must migrate to the AFA and remove the old array non-disruptively.

      For more information, see KB-000277.

    • To add an HFA or SFA to an existing AFA group, you must migrate from the HFA or SFA to AFA, and remove the HFA or SFA array.
    • This process is disruptive to host I/O and not recommended for iSCSI arrays which will undergo a discovery IP change.
    • Deduplication Domains cannot be merged, see Domains for details.

Verify the following factors:

  • Both arrays are in an active or standby state and are running the same array OS version.

    If not, update one or both arrays. To learn more about updating the array OS, refer to the Installation Guide or Hardware Guide for your array model.

  • Switches and inter-switch links must have their MTU's set to a value.

  • One or more subnets must be set to include group traffic in the Traffic Assignment menu. For any subnet where Allow Group is set to Yes, the switch ports connected to those array interfaces must be in the same Native VLAN. Arrays must have the same MTU setting on configured subnets.
  • Verify that the following factors are the same for both groups:
    • The data and management subnet
    • The data protocol

      The data ports must be connected.

      NOTE:

      A data subnet is required so the management subnet must be modified for use for management and data. The management and data ports must be in the same VLAN (L2 network, broadcast domain) as the other interfaces in that subnet.

      Data IPs must be added for group traffic between the arrays.

    • The L2 network or broadcast domain

      Both groups must also be in the same VLAN, on the same switch or switch-stack, or on switches connected via an inter-switch link or trunk.

    • The replication throttle levels

      Group-level throttles from the source group are automatically discarded and the destination group throttles are automatically applied.

      If the throttle levels are set differently on the two arrays that are associated with the destination and source groups, adjust the throttle level on the destination group based on your environment and requirements.

      The joining group inherits the destination group settings. You might see replication delays for the joining group if the throttle was set higher than its destination group. If the destination group has no throttle then the joining group no longer uses a throttle

    • When accessing the arrays using HPE CSI Driver for Kubenetes, make sure that the names of the tenant users are not the same on both arrays. In this configuration, the merge will fail if there is a user in both arrays with the same name.

Complete the following tasks if applicable:

  • Configure source and destination groups to have matching subnets in the Active network configuration.
    • Delete any draft network configs from source and destination groups.
    • On source and destination groups, “Edit” and “Save” the network configuration in order to over-write the “backup” network configuration.
  • Configure the joining group (the group being merged) as a single-array configuration.
    • You can merge only two groups at a time. Make sure that the joining group consists of a single array.
  • Perform the following replication tasks if you intend to merge replication partners into the same group:
    • Stop all replication and group merge processes in your environment.
    • Remove the hosting and joining groups from any replication configurations.
  • Unregister the HPE vCenter plugin, if you previously registered it.
    • Unregister the plugin at the source (joining) array because the destination group will service the plugin under its group management services. This does not cause downtime.
    • To learn more about unregistering the plugin, see the VMware Integration Guide. If you are using vVols, refer to the "Supported Features" section to determine if your vVol environment is impacted by the group merge.
  • Set offline all volumes associated with the joining group.
    NOTE: It is recommended that you perform this task using the GUI.

For Fibre Channel arrays:

  • After the volumes are offline, then offline all Fibre Channel interfaces on the source group. These tasks are easier using the GUI.
  • A data subnet is still required
    • The management subnet can be modified to be used for Mgmt+Data
    • Alternately, an unused Ethernet interface can be configured as a “data” interface, and a data IP added for group traffic between the arrays.
  • An alias is a human-friendly name associated with a WWPN. An alias may be defined on the switch (as a fabric-assigned alias) or using the array OS (as a user-assigned alias). For any two disjoint groups with user-assigned aliases, two potentially problematic situations can occur:
    • Different initiators with the same WWPN must use the same alias for that WWPN
    • Different initiators cannot be using the same alias for different WWPNs
    • In either case, you are presented with this information on the validation screen, and must manually perform the rename operation.