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 Mistakes

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.

Thanks to share!

Thanks to share!

Thanks to share!

Hi,

If the HCI appears normal but a single VM is experiencing intermittent network connectivity, troubleshooting should begin at the VM and virtual network levels, rather than immediately assuming there is an issue with the HCI cluster.

Things to check: VM Virtual NIC (vNIC); Virtual Switch / Network; User OS; Physical Network; HCI Monitoring & Logging

Troubleshooting sequence:
VM → vNIC → Virtual Switch/VLAN → Physical NIC/Uplink → Physical Switch.

If only one VM is experiencing issues while the other VMs and the HCI are functioning normally, the problem most likely lies with the vNIC, the guest OS, or the VM's network configuration.

"Intermittent"

I would like to add the line ----   Don't ask "where is it broken?" first -ask "when does it break, and what else happened at that exact time?"

You also need to add a methodology that leads with time-correlation and event-log review, not just a path walk.

Thanks to share!

Thanks to share