1/4
Muhammad Abid Lv3  Posted 2026-Aug-25 19:31
  
The key point is: you should not normally select one physical node’s IP as the replication IP when the Primary site has two nodes.

For your topology:

Primary Site: Node 1 + Node 2
DR Site: Node 3
Replication IP: 1 IP at Primary + 1 IP at DR

The preferred design is to use a virtual/cluster IP (VIP) at the Primary site for replication. That VIP should be able to move between Node 1 and Node 2 if one node fails.

So the communication would look like:

Primary Cluster VIP → DR Node IP

If your replication product specifically requires a replication IP configured on each node, then you would need:

Node 1 → Replication IP 1
Node 2 → Replication IP 2
DR Node → Replication IP

With only one Primary-site IP, the correct approach depends on whether your replication software supports a cluster/virtual replication IP.

If you tell me the exact product/technology you're using for replication, I can give you the exact configuration for your 2-node Primary → 1-node DR setup.
Newbie A4 Lv2  Posted 2026-Aug-28 00:39
  
Hello

Yes, the Secondary/DR site only has one node, while the Primary site has two nodes with a single IP address that can be used as the replication endpoint.
Depending on the particular Sangfor features and version, a Virtual IP (VIP) or an alternative replication architecture is needed because the Primary site only has one physical IP address. However, because the replication service could need each physical node to have its own accessible address, using a VIP by itself might not always fix the inter-node replication problem.
Korchai Lv2  Posted 2026-Aug-28 11:06
  
Assume the IP layout looks like this. Two primary nodes and one DR node.
The NOC team has supplied one IP address from each site with the same subnet and vlan ID.

Primary: 10.10.50.50; VLAN ID: 50
VLAN ID: 50 and DR: 10.20.50.50

Does this function? GW 10.10.50.1 10.20.50.1 Primary Site DR Site Node1 10.10.50.3 10.20.50.3 Node2 10.10.50.4 Cluster 10.10.50.2 SCP 10.10.50.6 10.20.50.6
Arsalan Israr Lv2  Posted 2026-Sep-06 12:04
  
Since Sangfor HCI replication uses a single gateway/egress point (not per-node IPs), you don't need separate IPs for each node. Two options:

Option 1 (Simple): Assign the one available IP to a single node's Edge interface and use that as your local replication IP. The cluster's internal overlay network will route the other node's data through this node automatically. Downside: if that node goes down, replication breaks (single point of failure).
Option 2 (Better): If your HCI version supports it, configure the Edge interface as a floating/virtual IP (VIP) shared between both nodes. If the active node fails, the IP fails over to the other node automatically, keeping replication running.
For production DR, Option 2 is recommended to avoid a single point of failure. Let me know your exact HCI version and I can point you to the specific config path.

I Can Help:

Change

Board Leaders

George ...

Weekly Sharers

George ...

Weekly Questioners