Reverting a Rolling Upgrade

To revert an upgrade, the Rolling Upgrade process is performed in reverse order by completing the following steps:

  1. Determine which partition is currently selected on each node by entering the show slot detail command.
    The include keyword can be used to simplify the output. In the following example, the Backup and both Standy nodes have been rebooted and are running the newer software version:
    # show slot detail | include "information|Image|State|ver:"
    Slot-1 information:
         State:               Operational
         Current State:       MASTER
         Image Selected:      primary
         Image Booted:        secondary
         Primary ver:         31.3.1.22
         Secondary ver:       31.3.0.106
    Slot-2 information:
         State:               Operational
         Current State:       BACKUP (In Sync)
         Image Selected:      primary
         Image Booted:        primary
         Primary ver:         31.3.1.22
         Secondary ver:       31.3.0.106
    Slot-3 information:
         State:               Operational
         Current State:       STANDBY
         Image Selected:      primary
         Image Booted:        primary
         Primary ver:         31.3.1.22
         Secondary ver:       31.3.0.106
    Slot-4 information:
         State:               Operational
         Current State:       STANDBY
         Image Selected:      primary
         Image Booted:        primary
         Primary ver:         31.3.1.22
         Secondary ver:       31.3.0.106
    
  2. Change the selected software partition to the old version on all nodes by entering the use image secondary command.
  3. When all nodes have the old software version selected, reboot each node running the new software version, starting with the Backup node first.
    When the Backup node reboots, it should go back to “In-Sync” with the Primary node.
    Note

    Note

    After each reboot, allow a few minutes for the network to converge before proceeding.
  4. Reboot any Standby nodes individually, ensuring that the rebooted node is operational before moving on to the next node.
When all nodes are running the old software version, the upgrade is reverted. At this point, all nodes should have both their selected and booted partitions set to the partition that contains the old software version.