Skip to main content
Solved

Connectivity problem

  • July 23, 2025
  • 3 replies
  • 74 views
S

Hi All,

I have been trying to setup a NC2 using AWS. The setup went through fine as shown below screenshot from the prism page:

Problem i see now is, the VM created (Ubuntu and Win10) are not been able to connect to internet for any updates or downloads (mirror urls). Also, we are not able to do ssh on those machines.

Any help is appreciated.

Best answer by dlink7

HI
Are your user vms on a different subnet than the CVM? They shouldn’t use the baremetal subnet.  I assume AWS native networking so you need to edit the UVM security group that was deployed with the cluster to allow ssh. You can send an email to dwayneATnutanix.com and I can setup a zoom.

This topic has been closed for replies.

3 replies

dlink7
Forum|alt.badge.img+19
  • dlink7
  • Moderator
  • 91 replies
  • Answer
  • July 23, 2025

HI
Are your user vms on a different subnet than the CVM? They shouldn’t use the baremetal subnet.  I assume AWS native networking so you need to edit the UVM security group that was deployed with the cluster to allow ssh. You can send an email to dwayneATnutanix.com and I can setup a zoom.

S
  • sharti
  • Author
  • Voyager
  • 1 reply
  • July 24, 2025

Thanks for your response. I have sent an email to dwayne@nutanix.com. Please let me know when you can setup the meeting. 

Jagoulden
Forum|alt.badge.img+2
  • Jagoulden
  • Trendsetter
  • 41 replies
  • July 30, 2025

Hi Sharti,

Thanks for the screenshot. I’m not sure if you have resolved the issue but here’s a suggestion: You’ve deployed Nutanix Cloud Clusters (NC2) on AWS, and Prism shows the cluster is healthy. However:

 

Current Issue:

  • VMs (Ubuntu and Win10) cannot access the internet (no updates/downloads).
  • SSH access to VMs fails.
  • CVMs are healthy, but guest VMs appear isolated.

 

Confirmed from Screenshot:

  • CVMs are distributed across hosts, healthy.
  • Ubuntu and Win10 VMs show low CPU usage, implying they are running but idle/network-bound.
  • No external network activity is visible.
  • This is a common Day 2 networking gap with NC2 on AWS.

 

Root Cause (Most Probable):

NC2 VMs do not get internet access by default unless:

  1. A NAT gateway or Internet Gateway (IGW) is attached to the subnet.
  2. Security Groups (SGs) allow:
    • Outbound egress (e.g., 0.0.0.0/0 on TCP 80/443).
    • Inbound SSH/RDP from your IP.
  3. AHV virtual NICs (vNICs) are correctly mapped to VPC subnet bridges.
  4. The default gateway and DNS inside the VM are set to AWS's VPC IPs.

 

What You Need to Check and Fix

1. Is There a NAT Gateway or IGW?

In AWS VPC:

  • Go to your VPC > Route Table.
  • Verify that the subnet where your guest VMs are running has:
  • Destination: 0.0.0.0/0  → Target: nat-xxxxxxxx or igw-xxxxxxxx

If not:

  • Add a NAT Gateway for private subnets or an IGW for public subnets.
  • Ensure the route is added to the route table.

 

2. Security Group Rules

For the ENIs (Elastic Network Interfaces) assigned to your AHV/VM NICs:

  • Egress (Outbound)
    • Allow all traffic: 0.0.0.0/0, TCP, UDP
  • Ingress (Inbound)
    • Allow SSH (TCP 22) and/or RDP (TCP 3389) from your public IP

Also check the AWS Network ACLs ensure they are not denying egress.

 

3. Prism Virtual Switch Mapping

In Prism (AHV networking view):

  • Ensure that the VM NICs are attached to a bridge that maps to the AWS subnet (usually br0 or custom bridge).
  • Verify that the uplink is active and bonded to the proper VPC ENI.

Use this on CVM to check:

manage_ovs show_bridges

 

4. Guest VM Default Gateway and DNS

Inside the VM:

  • Run ip a and ip r (or route -n) to check default route.
  • Run cat /etc/resolv.conf (Linux) or ipconfig /all (Windows) to verify DNS.

Make sure the default gateway is the VPC router IP (usually something like 10.128.0.1) and that DNS is either:

  • AWS default DNS
  • Or external (e.g., 8.8.8.8), if explicitly routed

 

Optional Test: Ping from CVM

From a Nutanix CVM:

ping google.com

curl -v https://mirror.centos.org/

If the CVM can reach the internet, but the guest VM cannot, it confirms the AHV-to-guest network bridge mapping is the problem.

 

Conclusion and Next Steps

  1. Validate your AWS networking (NAT/IGW, route table).
  2. Check AHV vSwitch mappings in Prism.
  3. Fix guest VM security group and network interface config.
  4. Test from inside VM and/or from CVM.

 

Terms & ConditionsAccessibility statement