Reputation: 7028
We have a special on-premise router in our office, which connects to a VPN. In this VPN there is a special IBM MQ server which we want to poll. The server provider demands the usage of a special client for polling which only works on windows. Since the rest of our IT-infrastructure is @AWS, we want to have the windows machine, which is supposed to poll the queues, as an EC2 instance.
To enable the connection our idea was that we set up a Raspberry Pi in our office, which connects via OpenVPN to a Client VPN Endpoint. The traffic from the Raspi is getting routed into the subnet, where the EC2 instance lives. All the traffic that the Raspi is receiving on a specific port, gets forwarded to the in-house router.
Unfortunately, we are not able to send anything from the EC2 instance to the Raspi. Is it even possible to route traffic from a subnet back to a Client VPN Endpoint?
Upvotes: 1
Views: 441
Reputation: 882
It's entirely possible; this problem is probably in your subnetting and routing.
Ensure that there is no overlap between the subnet CIDRs in your VPC, the VPN and the on-prem network.
Ensure the EC2 instance has an entry in its routing table that routes traffic for the on-prem network back via the client VPN endpoint. You can do this at VPC level by configuring the VPC's routing tables, so it applies to all instances in the VPC.
Ensure the Pi has IP forwarding enabled (echo 1 | sudo tee /proc/sys/net/ipv4/ip_forward
)
Ensure the Pi doesn't firewalling forwarded traffic (iptables-save
or iptables -L -v
)
Ensure the Pi has a route to get traffic up to the VPC, though it sounds like you have this in place already.
Upvotes: 2