Tag: Windows Server

  • Fixing “Keyset Does Not Exist” When Adding a Node to a Windows Server Failover Cluster

    Recently, I ran into a strange issue while trying to add newly built Windows Server 2022 hosts to an existing Microsoft Failover Cluster. Every attempt to add a node failed with the following error message:

    The server ‘host.zhengwu.org’ could not be added to the cluster.
    An error occurred while adding node ‘host.zhengwu.org’ to cluster ‘cluster.zhengwu.org’.

    Keyset does not exist

    At first glance, this error didn’t provide much direction. However, after some investigation, I discovered that the problem was caused by a mismatch in the node status between Microsoft Failover Cluster Manager and System Center Virtual Machine Manager (SCVMM).

    Root Cause: Node Status Out of Sync Between SCVMM and Failover Cluster

    One of the existing cluster nodes was still marked as Maintenance Mode in SCVMM, even though Failover Cluster Manager showed it as Online.

    This mismatch caused several issues:

    • The node appeared online in Failover Cluster Manager
    • Virtual machines were running on it normally
    • But SCVMM showed no VMs on the host
    • And SCVMM treated the host as if it were not fully operational

    Because SCVMM maintains cluster metadata and permissions, the inconsistent state led to the “Keyset does not exist” error when adding a new node.

    The following Windows Event log is also observed in Microsoft-Windows-FailoverClustering-Client/Diagnostic:

    AddClusterNodeEx (4570): Exception – error -2146893802 while adding node xxxx: Failed to Get Cluster Secret via CPrep Server on node xxxx

    Solution: Rejoin the Problematic Cluster Node to the Cluster

    To resolve the issue, I simply shut down the problematic cluster node.

    1. Go to the Failover Cluster Manager.
    2. Go to Nodes.
    3. Select the problematic node.
    4. Click More Actions in the right-side panel.
    5. Select Evict.
    6. I suggest reboot the node after evicting.
    7. Add the node back to the cluster after rebooting.