3.4. Upgrading Hosts in a Cluster from Red Hat Enterprise Linux 6 to Red Hat Enterprise Linux 7

To take advantage of all the features in Red Hat Enterprise Virtualization 3.6, upgrade the cluster compatibility version of your cluster from 3.5 to 3.6. All hosts in the cluster must be upgraded to Red Hat Enterprise Linux 7-based hypervisors before the upgrade of the cluster compatibility version is possible. For more information about the different hypervisor hosts see Introduction to Hypervisor Hosts in the Installation Guide.

Important

Red Hat recommends installing Next Generation Red Hat Virtualization Host (RHVH) 3.6 in place of Red Hat Enterprise Virtualization Hypervisor (RHEV-H) because it provides an upgrade path to the most recent version of Red Hat Virtualization 4.x through the Manager. Installations that use RHEV-H require a full reinstall of the hypervisors to upgrade to the most recent version of Red Hat Virtualization 4.x.

Note

This procedure assumes the current cluster compatibility version is 3.5, and is intended for clusters where virtual machines cannot be shut down during the upgrade. If a Red Hat Enterprise Linux 7-only cluster exists in the environment, shut down and move virtual machines that do not need to stay up to the Red Hat Enterprise Linux 7 cluster.
To enable the upgrade, the Manager must be configured to allow Red Hat Enterprise Linux 6 hosts and Red Hat Enterprise Linux 7 hosts to run in the same cluster. The InClusterUpgrade scheduling policy can be applied to your cluster to enable this functionality. See Scheduling Policies in the Administration Guide for more information about scheduling policies.
Before the the scheduling policy can be applied to the cluster, the following prerequisites apply:
  • No suspended virtual machines in the cluster.
  • No CPU pinning on any virtual machine in the cluster.
  • No NUMA pinning on any virtual machine in the cluster.
  • No virtual machines with the migration option Do not allow migration or Allow manual migration only selected.

Note

Virtual machines with Do not allow migration or Allow manual migration only selected, or with CPU pinning enabled must be shut down before these settings can be updated.
The following limitations apply during the upgrade process:
  • Virtual machines cannot be suspended.
  • High availability reservation for virtual machines is disabled. The Manager will still attempt to restart highly available virtual machines if they are failing.
  • Affinity groups will not be respected. After the upgrade is complete, the Affinity Rules Enforcement Manager will re-establish them.
  • Load balancing will not occur.
  • Migrating a virtual machine back to an older operating system version is not possible during upgrade. To run a virtual machine on an older host operating system, the virtual machine has to be stopped first.

Note

When placing a host into maintenance mode, all virtual machines on the host will be automatically migrated to hosts with an equal or greater operating system version.

Procedure 3.2. Upgrading Hosts in a Cluster from Red Hat Enterprise Linux 6 to Red Hat Enterprise Linux 7

  1. Enable the upgrade mode to allow the InClusterUpgrade policy to be set.
    # engine-config -s CheckMixedRhelVersions=false --cver=3.5
    
  2. Restart the engine.
    # service ovirt-engine restart
    
  3. In the Administration Portal, set the InClusterUpgrade scheduling policy on the cluster:
    1. Click the Clusters tab.
    2. Select the cluster and click Edit.
    3. Click the Scheduling Policy tab.
    4. From the Select Policy drop-down list, select InClusterUpgrade.
    5. Click OK.

    Note

    This policy must only be used during the upgrade process since it affects the way virtual machines are migrated.
  4. Reinstall each host with a hypervisor based on Red Hat Enterprise Linux 7:

    Note

    If required, migrate virtual machines manually as a precaution. If a host in the cluster has already been upgraded, then virtual machines will migrate to the Red Hat Enterprise Linux 7 host.

    Important

    Red Hat Enterprise Virtualization does not support using the Reinstall button to install a different hypervisor type, for example Red Hat Enterprise Linux host to Red Hat Virtualization Host. To install a different hypervisor type you must remove the host from the environment, install the new hypervisor, and then add it back.
    • For Red Hat Enterprise Linux Hosts:
      1. In the Administration Portal, click the Hosts tab and select the host to be upgraded.
      2. Click Maintenance to place the host into maintenance mode.
      3. To install a Red Hat Enterprise Linux 7 operating system on the host, see the Red Hat Enterprise Linux 7 Installation Guide.
      4. Subscribe the host to the required entitlements. See Subscribing to the Required Entitlements in the Installation Guide.
    • For RHEV-H or Next Generation RHVH:
      1. In the Administration Portal, click the Hosts tab and select the host to be upgraded.
      2. Click Maintenance to place the host into maintenance mode.
      3. Install the hypervisor:
        • To install Red Hat Virtualization Host 3.6 on the host, see Installing Red Hat Virtualization Host in the Red Hat Enterprise Virtualization Installation Guide.
        • To install Red Hat Enterprise Virtualization Hypervisor 7 on the host, see Installing the Red Hat Enterprise Virtualization Hypervisor in the Red Hat Enterprise Virtualization Installation Guide.

          Note

          A RHEV-H upgrade between major versions is not supported via the Manager. The installation must be done using installation media or via a PXE server on the host machine.
  5. Update the host's new SSH fingerprint in the Administration Portal to allow the Manager to connect to the host after reinstallation:
    1. Click the Hosts tab and select the upgraded host.
    2. Click Edit.
    3. Click the General tab and click Advanced Parameters.
    4. Click the fetch link to retrieve the new SSH fingerprint.
    5. Click OK.
  6. Install the Red Hat Enterprise Virtualization packages on the host:
    1. Click the Hosts tab and select the upgraded host.
    2. Click Reinstall to open the Install Host window.
    3. Click OK to reinstall the host.
  7. Remove the InClusterUpgrade scheduling policy:
    1. In the Administration Portal, click the Clusters tab.
    2. Select the cluster and click Edit.
    3. Click the Scheduling Policy tab.
    4. From the Select Policy drop-down list, select None. Alternatively apply a different scheduling policy.
    5. Click OK.
  8. Disable the upgrade mode if there are no more clusters to upgrade.
    # engine-config -s CheckMixedRhelVersions=true --cver=3.5
    
  9. Restart the engine.
    # service ovirt-engine restart
    
After you have upgraded all hosts, see Chapter 9, Post-Upgrade Tasks to complete the post-upgrade tasks.