Troubleshooting Mindset Series #015 When a VM Loses Network Connectivity but HCI Looks Normal — Where Should We Look?

One of our VMs suddenly experienced intermittent network connectivity.

From the HCI dashboard:

• CPU usage looked normal.
  
  
• Memory usage was normal.
  
  
• Storage status showed no obvious critical issue.
  
  
• The VM itself was still running normally.
  
  

However, users reported that the application hosted on the VM was occasionally unreachable.

This raised an important troubleshooting question:

If the VM and HCI resources look healthy, where should we look next?

Troubleshooting Approach

Instead of immediately restarting the VM, I prefer to troubleshoot from the network path.

A simple approach is:

1. Check VM network configuration

Verify:

• Virtual NIC is connected.
  
  
• Correct virtual switch/network is assigned.
  
  
• VLAN configuration is correct.
  
  
• No unexpected changes were made to the VM network settings.
  
  

2. Check connectivity from the VM

Test the communication step-by-step:

VM → Gateway → Other Server → Application

This helps identify where the connectivity actually breaks.

3. Check the physical network path

If the VM configuration looks correct, move one layer down:

• HCI host uplink
  
  
• Physical switch port
  
  
• VLAN
  
  
• Link status/errors
  
  
• Network teaming/bonding status
  
  

4. Compare with other VMs

This is an important isolation step.

If multiple VMs on the same host/network experience similar symptoms, the problem is less likely to be inside a single VM.

Common Mistake

A common troubleshooting mistake is to assume:

“The HCI dashboard is green, so the network must be fine.”

But infrastructure health and application connectivity are not always the same thing.

A host can show normal CPU, memory and storage utilisation while a problem exists somewhere along the VM-to-switch network path.

Troubleshooting Mindset

Don't troubleshoot based only on what looks abnormal.

Instead, identify:

Where does the communication stop?

Once the failure point is isolated, the troubleshooting scope becomes much smaller.

VM → vSwitch → HCI NIC → Physical Switch → VLAN → Gateway → Destination

That is usually much more effective than restarting components and hoping the problem disappears.