Cluster Node Already Up (0x13C5) – Quick Fix & Root Cause

I've seen this error a bunch of times with Windows Server 2019 and 2022 failover clusters. You try to start a node that's clearly down, and the system says "nope, it's already up." Yeah, frustrating. Let me show you the fix that works 9 times out of 10.

The Real Fix

Don't waste time rebooting the whole cluster. What's actually happening here is the cluster service on that node lost its state information. The registry or cluster database still holds a flag saying the node is online, even though the node is physically offline or hung.

Open PowerShell as Admin on another node that's still working. Run:

Clear-ClusterNode -Name "YourNodeName" -Force

Then on the problem node itself, run:

Start-ClusterNode

That's it. The reason step 3 works is: Clear-ClusterNode removes the stale online state from the cluster database. It basically tells the cluster "forget what you think you know about this node." Then Start-ClusterNode starts fresh.

Why This Happens

The cluster service uses a heartbeat mechanism. Each node sends a signal every 2 seconds to say "I'm alive." If a node's heartbeat stops but the cluster database didn't get the update before the node went offline, you get this 0x13C5 error. The database thinks the node is up because it never received the "I'm going down" message.

Common real-world trigger: You force-shutdown a node (holding the power button) or a network hiccup happens during a planned restart. The node goes offline faster than the cluster can record it.

Less Common Variations

If the PowerShell command doesn't work, you might be dealing with a corrupted cluster log. In that case:

1. Stop the cluster service on the problem node: Stop-Service clussvc
2. Delete the cluster log file at C:\Windows\Cluster\cluster.log
3. Restart the service: Start-Service clussvc
4. Try Start-ClusterNode again

Another variation: The node is stuck in an "Evicted" state. Check with Get-ClusterNode. If it shows State: Evicted, you need to add it back to the cluster, not just start it. Run:

Add-ClusterNode -Name "YourNodeName"

This happens when a previous eviction didn't complete cleanly. The cluster still sees the node's ID but won't let it join.

When You Should Reboot

Skip the reboot unless you've tried the above and still get the error. Rebooting the whole cluster is overkill. But if you've done both the Clear-ClusterNode and the log delete, and the node still won't start, reboot that single node. Sometimes the cluster service itself gets into a bad state that only a fresh start fixes.

Prevention

To stop this from happening again:

• Gracefully stop nodes. Use Suspend-ClusterNode -Drain before shutting down. This tells the cluster you're taking the node offline on purpose.
• Check your network. A flaky NIC can cause heartbeat loss. Run the cluster validation test (Test-Cluster) to check for network issues.
• Keep cluster log size small. The default log file can grow to 300MB. Set it to 50MB in the cluster properties: (Get-Cluster).LogSize = 50. Smaller logs are less likely to corrupt.

That's the whole thing. The error code 0x13C5 is annoying but it's not a real hardware problem most of the time. It's just the cluster's memory being wrong. Clear it, start fresh, move on.