Chapter 3. Troubleshooting a Self-Hosted Engine Deployment

To confirm whether the self-hosted engine has already been deployed run hosted-engine --check-deployed. An error will only be displayed if the self-hosted engine has not been deployed.

3.1. Troubleshooting the Manager Virtual Machine

Check the status of the Manager virtual machine by running hosted-engine --vm-status.

Note

Any changes made to the Manager virtual machine will take about 20 seconds before they are reflected in the status command output.
Depending on the Engine status in the output, see the following suggestions to find or fix the issue.
Engine status: "health": "good", "vm": "up" "detail": "up"
  1. If the Manager virtual machine is up and running as normal, you will see the following output:
    --== Host 1 status ==--
    
    Status up-to-date              : True
    Hostname                       : hypervisor.example.com
    Host ID                        : 1
    Engine status                  : {"health": "good", "vm": "up", "detail": "up"}
    Score                          : 3400
    stopped                        : False
    Local maintenance              : False
    crc32                          : 99e57eba
    Host timestamp                 : 248542
  2. If the output is normal but you cannot connect to the Manager, check the network connection.
Engine status: "reason": "failed liveliness check", "health": "bad", "vm": "up", "detail": "up"
  1. If the health is bad and the vm is up, the HA services will try to restart the Manager virtual machine to get the Manager back. If it does not succeed within a few minutes, enable the global maintenance mode from the command line so that the hosts are no longer managed by the HA services.
    # hosted-engine --set-maintenance --mode=global
  2. Connect to the console. When prompted, enter the operating system's root password. For more console options, see https://access.redhat.com/solutions/2221461.
    # hosted-engine --console
  3. Ensure that the Manager virtual machine's operating system is running by logging in.
  4. Check the status of the ovirt-engine service:
    # systemctl status -l ovirt-engine
    # journalctl -u ovirt-engine
  5. Check the following logs: /var/log/messages, /var/log/ovirt-engine/engine.log and /var/log/ovirt-engine/server.log.
  6. After fixing the issue, reboot the Manager virtual machine manually from one of the self-hosted engine nodes:
    # hosted-engine --vm-shutdown
    # hosted-engine --vm-start

    Note

    When the self-hosted engine nodes are in global maintenance mode, the Manager virtual machine must be rebooted manually. If you try to reboot the Manager virtual machine by sending a reboot command from the command line, the Manager virtual machine will remain powered off. This is by design.
  7. On the Manager virtual machine, verify that the ovirt-engine service is up and running:
    # systemctl status ovirt-engine.service
  8. After ensuring the Manager virtual machine is up and running, close the console session and disable the maintenance mode to enable the HA services again:
    # hosted-engine --set-maintenance --mode=none
Engine status: "vm": "down", "health": "bad", "detail": "unknown", "reason": "vm not running on this host"
  1. If you have more than one host in your environment, ensure that another host is not currently trying to restart the Manager virtual machine.
  2. Ensure that you are not in global maintenance mode.
  3. Check the ovirt-ha-agent logs in /var/log/ovirt-hosted-engine-ha/agent.log.
  4. Try to reboot the Manager virtual machine manually from one of the self-hosted engine nodes:
    # hosted-engine --vm-shutdown
    # hosted-engine --vm-start
Engine status: "vm": "unknown", "health": "unknown", "detail": "unknown", "reason": "failed to getVmStats"
This status means that ovirt-ha-agent failed to get the virtual machine's details from VDSM.
  1. Check the VDSM logs in /var/log/vdsm/vdsm.log.
  2. Check the ovirt-ha-agent logs in /var/log/ovirt-hosted-engine-ha/agent.log.
Engine status: The self-hosted engine's configuration has not been retrieved from shared storage
If you receive the status The hosted engine configuration has not been retrieved from shared storage. Please ensure that ovirt-ha-agent is running and the storage server is reachable there is an issue with the ovirt-ha-agent service, or with the storage, or both.
  1. Check the status of ovirt-ha-agent on the host:
    # systemctl status -l ovirt-ha-agent
    # journalctl -u ovirt-ha-agent
  2. If the ovirt-ha-agent is down, restart it:
    # systemctl start ovirt-ha-agent
  3. Check the ovirt-ha-agent logs in /var/log/ovirt-hosted-engine-ha/agent.log.
  4. Check that you can ping the shared storage.
  5. Check whether the shared storage is mounted.

Additional Troubleshooting Commands:

Important

Contact the Red Hat Support Team if you feel you need to run any of these commands to troubleshoot your self-hosted engine environment.
  • hosted-engine --reinitialize-lockspace: This command is used when the sanlock lockspace is broken. Ensure that the global maintenance mode is enabled and that the Manager virtual machine is stopped before reinitializing the sanlock lockspaces.
  • hosted-engine --clean-metadata: Remove the metadata for a host's agent from the global status database. This makes all other hosts forget about this host. Ensure that the target host is down and that the global maintenance mode is enabled.
  • hosted-engine --check-liveliness: This command checks the liveliness page of the ovirt-engine service. You can also check by connecting to https://engine-fqdn/ovirt-engine/services/health/ in a web browser.
  • hosted-engine --connect-storage: This command instructs VDSM to prepare all storage connections needed for the host and and the Manager virtual machine. This is normally run in the back-end during the self-hosted engine deployment. Ensure that the global maintenance mode is enabled if you need to run this command to troubleshoot storage issues.