Troubleshooting and Support
Where to look first
Split the problem by layer. AWS layer: CloudFormation stack events, EC2 status checks, VPC route tables, security groups, NACLs, VPN/Direct Connect metrics. Nutanix layer: Prism alerts, NCC output, task history, CVM logs, and the support bundle. The NC2 console task log bridges both.
Common NC2 failures
- Deployment fails: insufficient IAM permissions, bare-metal quota, no capacity in the AZ, subnet IP exhaustion, or an unsupported instance type.
- Prism unreachable: security group missing 9440, route missing, EIP/NAT misconfigured, or Prism service down.
- Replication slow: VPN throughput, MTU, or replication schedule.
- Node not joining: time skew/NTP, network mismatch, or version mismatch.
Worked example
A new cluster deploys but Prism at 9440 times out from the corporate network. The operator confirms the instance is running and the CloudFormation stack is complete. The security group allows 9440 only from the VPC CIDR, not the corporate CIDR; the VPN route exists. Adding the corporate CIDR to the security group fixes access. If it still failed, the next checks are NACL ephemeral ports and Prism service health.
Traps
Opening a ticket without the support bundle; troubleshooting only at the Nutanix layer when the security group is wrong; changing multiple variables at once; assuming AWS Support covers Nutanix software or vice versa. Engage the correct support owner with the NC2 cluster ID and AWS case ID.