Cluster Creation Failed

[nrajkumarf3]

Hi Everyone, 

I am encountering an issue with Databricks cluster creation and would appreciate your support in resolving it.

During cluster launch, the creation fails with the following network health check error:

Error Details:

  - Error Code: X_NHC_CONTROL_PLANE_UNREACHABLE
  - Message: Instance failed network health check before bootstrapping with fatal error: X_NHC_CONTROL_PLANE_UNREACHABLE
  - Failed Components:
    1.  centralindia.azuredatabricks.net (Outcome: unreachable, curl: (28) Connection timed out after 10001 milliseconds)
    2.  www.databricks.com (Outcome: unreachable, curl: (28) Failed to connect to www.databricks.com port 443: Connection timed out)

Could you please assist in checking the network connectivity and control plane reachability for this workspace to resolve this issue?

Please let me know if you need any additional logs or details from my end.

[ThomazNeto]

Hi,

Both checks timing out, the control plane (centralindia.azuredatabricks.net) and www.databricks.com, usually means the cluster subnets have no outbound path at all, not a problem with one specific endpoint. The error codes doc says the same: when multiple components fail together, the root cause is usually shared, like a DNS or firewall change.

Things to check, in this order:

1. Outbound connectivity on the subnets. If your workspace uses VNet injection on a VNet created after March 31, 2026, Azure creates it private by default, with no outbound internet access. The secure cluster connectivity docs say new workspaces after that date need an explicit outbound method, and recommend an Azure NAT gateway attached to both workspace subnets. If you use the default managed VNet, Azure Databricks creates the NAT gateway for you, so skip to the next items.

2. NSG and route tables. Check the NSG outbound rules on both Databricks subnets. If you have a UDR sending traffic to a firewall, make sure the firewall allows outbound HTTPS to the Azure Databricks control plane and that the Azure Databricks service tags route to the internet.

3. DNS. If the VNet uses a custom DNS server, test temporarily with Azure DNS (168.63.129.16) to rule it out.

4. Test from inside the network. Spin up a small VM in the same subnet and run curl against https://centralindia.azuredatabricks.net. If it times out there too, the issue is confirmed to be in the VNet configuration, not in Databricks.

If all of that checks out and it still fails, the doc's guidance is to open a support case.

Docs:
Classic compute termination error codes (network health check section): https://learn.microsoft.com/en-us/azure/databricks/compute/troubleshooting/cluster-error-codes
Secure cluster connectivity and NAT gateway: https://learn.microsoft.com/azure/databricks/security/secure-cluster-connectivity