You can manually upgrade hosts in a host upgrade unit group.

Prerequisites

Verify that the upgrade coordinator is updated. See Upgrade the Upgrade Coordinator.

Procedure

  1. In the upgrade coordinator, navigate to the Host Upgrade tab.
  2. Select an enabled host upgrade unit group.
  3. Select Actions > Change State > Disabled.
    If you have other enabled host upgrade unit groups, set them to Disabled.
  4. Click Start to perform the pre-upgrade process.
  5. Wait for the host upgrade to pause.
  6. Upgrade your ESXi host manually.
    Note: If a host fails during the upgrade, reboot the host and try the upgrade again.
    1. Put the ESXi host in Maintenance mode.
    2. Navigate to the ESXi offline bundle location from the NSX Manager.
      http://<nsx-manager-ip-address>:8080/repository/<target-nsx-t-version>/metadata/manifest.
    3. Download the ESXi offline bundle to /tmp on ESXi.
    4. Upgrade the ESXi host.
      esxcli software vib install -d /tmp/<offline-bundle-name>.
  7. Upgrade your KVM host manually.
    Note: If a host fails during the upgrade, reboot the host and try the upgrade again.
    1. Download the upgrade script.
      http://<nsx-manager-ip-address>:8080/repository/<target-nsx-t-version>/HostComponents/<os-type>/upgrade.sh

      Where the os_type is rhel76_x86_64, rhel77_x86_64, xenial_amd64, linux64-bionic, linux64-sles12sp3, or linux64-sles12sp4.

    2. Upgrade the KVM host.
      upgrade.sh <host-upgrade-bundle-url> <checksum>

      Where the host upgrade bundle URL is, http://<nsx-manager-ip-address>:8080/xyz where xyz, is one of the paths from the http://<nsx-manager-ip-address>:8080/repository/<target-nsx-version>/metadata/manifest file.

      For example, http://<nsx-manager-ip-address>:8080/repository/3.0.0.0.0.99999999/HostComponents/rhel76_x86_64/nsx-lcp-2.3.0.0.0.9999999-rhel76_x86_64.tar.gz.
      To retrieve checksum, log in to NSX Manager as a root user and run the following command: 
      sha256sum /repository/<target-nsx-version>/HostComponents/<os-type>/*
  8. In the upgrade coordinator, navigate to the Hosts tab and refresh the page.
    All the manually upgraded hosts appear in the upgraded state.
  9. After the upgrade is successful, verify that the latest version of NSX-T Data Center packages is installed on the vSphere and KVM hosts.
    • For vSphere hosts, enter esxcli software vib list | grep nsx.
    • For Ubuntu hosts, enter dpkg -l | grep nsx.
    • For SUSE Linux Enterprise Server, Red Hat, or CentOS hosts, enter rpm -qa | egrep 'nsx|openvswitch|nicira'.
  10. Power on the tenant VMs of standalone ESXi hosts that were powered off before the upgrade.
  11. Migrate the tenant VMs of managed ESXi hosts that are part of the DRS disabled cluster to the appropriate host.
  12. Power on or return the tenant VMs of ESXi hosts that are part of a DRS disabled cluster that were powered off before the upgrade.
  13. (Optional) In the NSX Manager appliance, select System > Appliances and verify that all the status indicators for host and transport node deployment appear as installed and connection status is up and green.
  14. In the upgrade coordinator, navigate to the Hosts tab and select a disabled host upgrade unit group.
  15. Select Actions > Change State > Enabled.
    If you have other disabled host upgrade unit groups, set them to Enabled.

What to do next

You can proceed with the upgrade only after the upgrade process finishes successfully. See Upgrade Management Plane.

If there are upgrade errors, you must resolve the errors. See Troubleshooting Upgrade Failures.