VMware Networking Community
algoeit
Contributor
Contributor
Jump to solution

NSX and Stretched VSAN problem.

We are building a new enviroment where we have a streched vsan between two PoDs. We have one vCenter that manages all hosts in the two different L3 subnets.

The problem is now that when we shutdown the Pod where vCenter, NSX manager, PSC and Universal Controll cluster is running, non of the components failover since they have there own subnet. There is no Layer2 streching possible only within an VXLAN.

So two things are happening. Compute VM's will failover to the other POD because of HA. VXLAN is there because of CDO is enabled, but since there is no control plane there is no North-South traffic between the ULDR and the edge.

Streched VSAN and HA are still working since this is not needed for vCenter.

The only solution I now can think of is to have a seperate vcenter/psc/nsx mgr/controll cluster inside a VXLAN. But this is like a cause-effect problem.

Hope this is clear and maybe somebodyhas come across this scenario already.

Reply
0 Kudos
1 Solution

Accepted Solutions
algoeit
Contributor
Contributor
Jump to solution

We eventually create an VPN beteen sites to achieve L2. So we could go on with a streched cluster between the sites.

In case of a datacenter failure the gateway of this segment only needs to be changed.

View solution in original post

Reply
0 Kudos
3 Replies
algoeit
Contributor
Contributor
Jump to solution

Also the control VM of the ULDR is not available since it has an IP from only one POD.

Reply
0 Kudos
lhoffer
VMware Employee
VMware Employee
Jump to solution

Check out the Control VM Failure & Recovery section on page 132 of the VMware® NSX for vSphere Network Virtualization Design Guide ver 3.0

At a high level, floating static routes on the ESGs for the subnets behind the DLR will allow ingress traffic to keep flowing even after the routing adjacencies are lost.

Reply
0 Kudos
algoeit
Contributor
Contributor
Jump to solution

We eventually create an VPN beteen sites to achieve L2. So we could go on with a streched cluster between the sites.

In case of a datacenter failure the gateway of this segment only needs to be changed.

Reply
0 Kudos