Issue with DNS resolution in Holodeck dual-site deployment

[chinthub]

Describe the bug

I am facing a DNS-related issue in a dual-site deployment.

Deployment completes successfully without errors

Network design
Site A subnet: 10.1.0.0/16

Site B subnet: 10.2.0.0/16

DNS configuration (from Holodeck Router)
When I run:

Get-HoloDeckDNSConfig

The output shows:

Site A DNS: 10.1.168.129

Site B DNS: 10.2.168.129

This matches the expected Holodeck DNS design I believe.

What I am seeing on the ESXi hosts
After deployment:

Site A ESXi hosts

IPs assigned: 10.1.16.101, 10.1.16.102, etc.

Default gateway: 10.1.16.1

DNS configured as: 10.1.16.1 (same as gateway)

Site B ESXi hosts

IPs assigned: 10.2.16.101, 10.2.16.102, 10.2.16.103, etc.

Default gateway: 10.2.16.1

DNS configured as: 10.2.16.1

However, the DNS should be:

Site A: 10.1.168.129

Site B: 10.2.168.129

Troubleshooting done so far
From the ESXi DCUI, I tested "Test Management Network"

DNS resolution fails on both sites

I manually changed DNS on a few ESXi hosts in both sites:

From gateway IP → to 10.1.168.129 / 10.2.168.129

Even after manual change:

10.1.168.129 is not reachable

DNS still does not resolve

This suggests the DNS service itself is either unreachable or not correctly routed. Please advise

Reproduction steps

1. Deploy a Holodeck dual-site environment in Provision-only mode with Site A 10.1.0.0/16 and Site B 10.2.0.0/16.
2. Complete the deployment and retrieve DNS details using get-holodeck-dns-config.
3. Observe DNS IPs as 10.1.168.129 (Site A) and 10.2.168.129 (Site B).
4. Verify ESXi host network settings and note DNS configured as the subnet gateway (10.1.16.1 / 10.2.16.1).
5. From ESXi DCUI, run Test Management Network and confirm DNS resolution fails and the Holodeck DNS IP is unreachable.

Expected behavior

The ESXi hosts in Site A and Site B should automatically receive the correct DNS servers (10.1.168.129 and 10.2.168.129) as defined in the Holodeck DNS configuration. DNS resolution from the ESXi management network should work successfully without any manual DNS or routing changes after deployment.

Additional context

No response

[bensier]

We actually use /20 for the networks.. I'm not sure what specifying a /16 would do, it has not been tested. (Well now I guess we have an idea!)

If there is a reason you'd need that many more IP's and potential subnets we could take a look at it but most things are mapped to specific VLAN's and subnets are carved out of that /20 in a prescriptive way.

I see in the dual-site section of the docs it's not clear that you should use a /20. We will update this. Above that when you look at the CIDR parameter for New-HoloDeckInstance it states:

'If you wish to use a custom CIDR, provide a CIDR of /20 size. '

Apologies for the lack of clarity.

[chinthub]

Thanks for the response. I will try with /20 and update here.

[chinthub]

Installed successfully without any issue, but still DNS check fail when performing "Test Management Network" from ESXi DCUI.

[dhruv-tyagi-broadcom]

Can you share the steps you followed? Did you make use of the same Holorouter or used a new one?

[chinthub]

I always redeploy new holorouter. These are the commands I ran.

HoloDeck Commands for Dual Site (Provision Only):

New-HoloDeckConfig -TargetHost '192.168.77.18' -UserName root -Password VMware123!

Import-HoloDeckConfig -ConfigID xxxx

New-HoloDeckNetworkConfig -Site a

New-HoloDeckNetworkConfig -Site b

Set-HoloRouter -dualsite

New-HoloDeckInstance -Site a -Version 9.0.1.0 -ProvisionOnly -DepotType Offline

New-HoloDeckInstance -Site b -Version 9.0.1.0 -ProvisionOnly -DepotType Offline (ran parrellely on another putty session after triggering Site A completely)