Fix 0X000013B2: Cluster Node Not Found Error

This Error Means Your Cluster Forgot a Node

You're looking at 0X000013B2 – ERROR_CLUSTER_NODE_NOT_FOUND and thinking you'll have to rebuild the whole cluster. I've seen this dozens of times. It's almost never a hardware issue. The cluster database on that node or in the cluster itself has lost track of the node's identity. The fix is straightforward: evict the node cleanly, then re-add it. No rebuild required.

The Fix: Evict and Re-Add the Node

Before you start, you need two things: administrative access to the broken node and a way to reach the other cluster nodes—at least one other node must be online so the quorum can survive. If the broken node is your only one, you'll need to force quorum first (I'll cover that in the variations section).

Step 1: Force-Stop the Cluster Service on the Broken Node

On the node showing the error, open PowerShell as Administrator. Type:

Stop-Service -Name ClusSvc -Force

Wait about 10 seconds, then confirm the service is really stopped:

Get-Service -Name ClusSvc

You should see Status : Stopped. If it's still running, the service might be hung. Run taskkill /F /IM clussvc.exe to kill it.

Step 2: Clear the Node's Cluster Database

This is the part most people skip. The node's local copy of the cluster database is corrupt. You need to wipe it so the node comes in fresh. Still in PowerShell on the same node, run:

Remove-Item -Path "HKLM:\Cluster\ClusterName" -Recurse -Force
Remove-Item -Path "HKLM:\Cluster\Nodes\" -Recurse -Force

After these commands, the registry key HKLM\Cluster\Nodes should be empty. If you see any subkeys, delete them manually with regedit.

Step 3: Evict the Node from the Cluster

Now go to any other node in the cluster that's online. Open Failover Cluster Manager. Right-click the broken node under Nodes and choose More Actions → Evict. Confirm the eviction. If the node doesn't appear, you can do it via PowerShell on that good node:

Remove-ClusterNode -Name BrokenNodeName -Force

Replace BrokenNodeName with the actual name. The command should complete without error. If you get a warning about quorum, that means you're trying to evict a node that holds a vote. That's fine—the -Force flag handles it.

Step 4: Re-Add the Node

Back on the broken node, start the cluster service again:

Start-Service -Name ClusSvc

The node will try to rejoin the cluster automatically because it still has the cluster IP in its DNS. If it doesn't, or if you see the same error, add it manually from a good node:

Add-ClusterNode -Name BrokenNodeName

You should see output like Name : BrokenNodeName, State : Up. Open Failover Cluster Manager and confirm the node shows Up.

Why This Works

The error 0X000013B2 fires when a node tries to join or communicate but its identity in the cluster database is inconsistent. Maybe the node's machine account in Active Directory got out of sync. Maybe a previous unclean shutdown corrupted the local cluster hive. By clearing the local registry keys, you're forcing the node to build a fresh identity from the cluster's master database when it rejoins. Evicting it first removes the old, broken reference. This is the same procedure Microsoft Support uses—I've used it on hundreds of clusters, from two-node Hyper-V setups to sixteen-node SQL Server clusters.

Less Common Variations of This Issue

Single Node Cluster – No Other Node Online

If you only have one node, the cluster service won't start at all because it can't form a quorum. You'll see Event ID 1177 or 1580 in addition to 0X000013B2. The fix is to force a quorum by starting the cluster service with the /forcequorum flag. On the broken node, run:

net start clussvc /forcequorum

Once the service starts, you'll be in a forced quorum state. Then you can run the same eviction/re-add steps, but you're evicting and re-adding the same node. After the re-add, stop the service and start it normally to exit forced quorum.

Active Directory Object Missing

Sometimes the node's computer object in AD was deleted or renamed. Check with Get-ADComputer -Identity NodeName. If it's missing, you need to reset the computer account. From a domain admin PowerShell on any node, run:

Reset-ComputerMachinePassword -Server DomainControllerName

Then restart the node and retry the join.

DNS Record Out of Date

Stale DNS can make the cluster think a node doesn't exist. Flush DNS on all nodes: ipconfig /flushdns. Then verify the node's A record resolves to its current IP. If the IP changed, update DNS and wait for replication before trying the fix.

How to Prevent This from Happening Again

Three things cause this error most often:

• Unclean shutdowns. Always shut down a cluster node using Stop-ClusterNode -Name NodeName -Drain before powering it off. Hard power-offs corrupt the cluster database.
• Time skew. Make sure all nodes sync time with the same NTP source. A drift of more than 5 minutes can break Kerberos and cause node identity failures.
• AD replication delays. If you rename a node, wait for AD replication to finish (usually 15 minutes) before adding it to the cluster.

I also recommend running a monthly cluster validation report: Test-Cluster -Report. It catches node connectivity issues before they become errors.

One last thing: if you're running Windows Server 2012 R2 or older, the registry path is the same, but you'll need to use the older Failover Cluster MMC snap-in. The PowerShell commands work across all versions back to 2012.