AlwayOn and Failover Cluster Manager Failover Properties

  • I ran across an article that left me wondering what the Failover properties are supposed to be set to regarding failovers for the availability group role in Failover Cluster Manager's Properties Failover tab. Specifically should it be left on the default of Allow failback immediately, or should it be set to Prevent Failback. The specific article that got me wondering was https://docs.microsoft.com/en-us/sql/sql-server/failover-clusters/windows/failover-policy-for-failover-cluster-instances?view=sql-server-2017 with the note:
    Important
    Automatic failovers to and from an FCI are not allowed in an Always On availability group. However, manual failovers to and from and FCI are allowed in an Always On availability group.

  • For any disruption at the primary node you want the service to failover to the secondary. However, it is better to keep the fail back to manual so you can review the issues causing disruption and manually failback at a quiet window without service disruption.

  • That note doesn't really apply to allowing or preventing  automatic failback. That note is referring to a limitation when running an AlwaysOn Availability Group with one or more Failover Clustered Instances.

    There is a more illuminating note about this in another article:

    SQL Server Failover Cluster Instances (FCIs) do not support automatic failover by availability groups, so any availability replica that is hosted by an FCI can only be configured for manual failover.

    Link here.

    What this means is that if you have an Availability Group with an FCI as a replica, the FCI node must be set to Manual failover, not Automatic. To the question about automatic failback - it doesn't work with AlwaysOn AG resources in the cluster, and for good reason. Failover Cluster Manager should never be used to failover AG resources because SQL Server has important tasks to run in the event of a failover that FCM does not know about or perform. For this reason, the AG automatically resets the preferred and possible owner settings in the cluster resource after a failover to prevent a failback from occurring.

    See here for more info.

Viewing 3 posts - 1 through 2 (of 2 total)

You must be logged in to reply to this topic. Login to reply