Navigation

Upgrade Issues

Failed Rolling Upgrade

Warning

Never attempt to upgrade Ops Manager without backing up the application database. In many recovery scenarios, a current backup is required to return your instance to health.

To recover from a failed rolling upgrade follow these steps in the following order:

  1. Restart your Ops Manager instances.

  2. Check to see if you have upgraded all Ops Manager hosts that support your Ops Manager instance.

    If Ops Manager detects two running versions, it remains in Upgrade Mode. Stop the Ops Manager Application on the host running an earlier version, upgrade that host, then restart that host.

  3. Restore the Application Database to a previous state after the upgrade process had started, but before it has completed.

    1. Copy of all the Ops Manager log files.
    2. Stop all Ops Manager instances including backup daemons.
    3. Restore the Application Database to a backup taken before the upgrade was started.
    4. Reinstall the previous version of Ops Manager.
    5. Restart all Ops Manager instances.
    6. Contact MongoDB Support and provide all details.
  4. Leave Ops Manager in Upgrade Mode and contact MongoDB Support.

    • If migrations cannot be applied, Ops Manager remains in Upgrade Mode.
    • If you can accept your Ops Manager instance in this state within the bounds of your SLA, contact MongoDB Support and provide details of the problem
←   Monitoring Host Issues  →