This document outlines the steps to set up a one tunnel IPSec Site to site VPN on AWS and a VM on another cloud provider (Packet) running Strongswan.
(References:
https://docs.aws.amazon.com/vpn/latest/s2svpn/SetUpVPNConnections.html
(This can be any cloud provider. It does not have to be Packet)
Ensure you already have your SSH key on Packet. Click on New Server.
Create an small server with Ubuntu 18.01 LTS, and associate your public SSH key with it.
# ssh into server
sudo apt update
sudo apt install strongswan # strongswan will be active(running) by default
# Version of strongswan in this case is 5.6.2-1
After this is done, we will refer to this VM henceforth as Packet VPN server.
On AWS, Select the VPC service from the list of Services.
Create a VPC (using a private IP address range e.g 172.17.0.0/16).
On the Virtual Private Network(VPN) section on the left, select Customer Gateways.
Click Create Customer Gateway, on Routing, select Static, then add the internet-routable IP address of the Packet VPN server.
Next select Virtual Private Gateways on the section on the left.
Click on Create Virtual Private Gateway (VPG), leave the Amazon default ASN checked. Then click Create Virtual Private Gateway.
Select the VPG you created, click Actions, then click Attach to VPC.
Attach to the VPC you created earlier.
Following the creation of your VPG, visit the Route Tables section under Virtual Private Cloud.
Select the routing table corresponding to your subnet(s). (In this case, select the one associated with your VPC)
Afterwards, click the Route Propagation tab and then select the vgw identifier for the virtual private gateway that was created earlier.
Click Edit route propagation to view the Propagate checkbox, click the checkbox and choose Save.
Update security group to allow access to instances in your VPC from your network. Enable SSH, RDP and ICMP access.
To do that, in the navigation pane, choose Security Groups under Security.
Select the default security group for the VPC and enable SSH, RDP and ICMP access.
Finally, visit the VPN section on the left, choose Site-to-Site VPN Connections.
Click Create VPN Connection.
In the dialog that results, select the virtual private gateway (vgw) and the customer gateway that you have previously created.
For routing options, select Static then enter the IP CIDR of the Packet VPN server (e.g 147.75.44.166/31) i.e enter all of the IP prefixes of your on-premises network.
Leave the Tunnel Options to be automatically generated by Amazon.
Then click Create VPN Connection.
After the VPN connection has been created, the State of the connection should switch to available in a few minutes.
As the connections have not been made yet to the VPN servers, it is perfectly normal for the icons to be red.
Select the VPN connection that was created, and then note the Tunnel 1 and Tunnel 2 IP addresses.
Click the Download Configuration button when finished.
In the dialog box, select Strongswan for Vendor, Platform Ubuntu 16.04 (this is all I saw), Software Strongswan 5.5.1+.
Then click Download.
Follow the instructions on the downloaded file from AWS.
I did the steps in IPSEC Tunnel #1, did not do setup for Tunnel #2.
I omitted steps #2 - #5 then moved on to the Automated Tunnel Healthcheck and Failover section of the document.
On Packet VPN server
root@kosy:/etc# ipsec status # to ensure both of your tunnels are ESTABLISHED
Security Associations (1 up, 0 connecting):
Tunnel1[1]: ESTABLISHED 56 seconds ago, 147.75.102.25[147.75.102.25]...35.156.127.214[35.156.127.214]
Tunnel1{1}: INSTALLED, TUNNEL, reqid 1, ESP in UDP SPIs: c59e6b8b_i 5ad4d103_o
Tunnel1{1}: 0.0.0.0/0 === 0.0.0.0/0
root@kosy:/etc# ip route # to ensure route table entries were created for each of your tunnel interfaces, and the destination is the default via 147.75.102.24 dev bond0 onlink
10.0.0.0/8 via 10.80.166.0 dev bond0
10.80.166.0/31 dev bond0 proto kernel scope link src 10.80.166.1
147.75.102.24/31 dev bond0 proto kernel scope link src 147.75.102.25
169.254.40.124/30 dev Tunnel1 proto kernel scope link src 169.254.40.126
172.17.0.0/16 dev Tunnel1 scope link metric 100
root@kosy:/etc# iptables -t mangle -L -n # to ensure entries were made for both of your tunnels in both the INPUT and FORWARD chain
Chain PREROUTING (policy ACCEPT)
target prot opt source destination
Chain INPUT (policy ACCEPT)
target prot opt source destination
MARK esp -- 35.156.127.214 147.75.102.25 MARK set 0x64
Chain FORWARD (policy ACCEPT)
target prot opt source destination
TCPMSS tcp -- 0.0.0.0/0 0.0.0.0/0 tcp flags:0x06/0x02 TCPMSS clamp to PMTU
Chain OUTPUT (policy ACCEPT)
target prot opt source destination
Chain POSTROUTING (policy ACCEPT)
target prot opt source destination
root@kosy:/etc# ifconfig # to ensure the correct 169.254.x addresses were assigned to each end of your peer-to-peer virtual tunnel interfaces
Tunnel1: flags=209<UP,POINTOPOINT,RUNNING,NOARP> mtu 1419
inet 169.254.40.126 netmask 255.255.255.252 destination 169.254.40.125
inet6 fe80::200:5efe:934b:6619 prefixlen 64 scopeid 0x20<link>
tunnel txqueuelen 1000 (IPIP Tunnel)
RX packets 0 bytes 0 (0.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 0 bytes 0 (0.0 B)
TX errors 6 dropped 0 overruns 0 carrier 6 collisions 0
bond0: flags=5187<UP,BROADCAST,RUNNING,MASTER,MULTICAST> mtu 1500
inet 147.75.102.25 netmask 255.255.255.254 broadcast 255.255.255.255
inet6 2604:1380:2001:2e00::1 prefixlen 127 scopeid 0x0<global>
inet6 fe80::ec4:7aff:fee5:48fe prefixlen 64 scopeid 0x20<link>
ether 0c:c4:7a:e5:48:fe txqueuelen 1000 (Ethernet)
RX packets 5110 bytes 3775495 (3.7 MB)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 7377 bytes 765096 (765.0 KB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
bond0:0: flags=5187<UP,BROADCAST,RUNNING,MASTER,MULTICAST> mtu 1500
inet 10.80.166.1 netmask 255.255.255.254 broadcast 255.255.255.255
ether 0c:c4:7a:e5:48:fe txqueuelen 1000 (Ethernet)
enp0s20f0: flags=6211<UP,BROADCAST,RUNNING,SLAVE,MULTICAST> mtu 1500
ether 0c:c4:7a:e5:48:fe txqueuelen 1000 (Ethernet)
RX packets 5110 bytes 3775495 (3.7 MB)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 7377 bytes 765096 (765.0 KB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
device memory 0xdf120000-df13ffff
enp0s20f1: flags=6147<UP,BROADCAST,SLAVE,MULTICAST> mtu 1500
ether 0c:c4:7a:e5:48:ff txqueuelen 1000 (Ethernet)
RX packets 0 bytes 0 (0.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 0 bytes 0 (0.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
device memory 0xdf100000-df11ffff
lo: flags=73<UP,LOOPBACK,RUNNING> mtu 65536
inet 127.0.0.1 netmask 255.0.0.0
inet6 ::1 prefixlen 128 scopeid 0x10<host>
loop txqueuelen 1000 (Local Loopback)
RX packets 0 bytes 0 (0.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 0 bytes 0 (0.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
On AWS VPN Site-to-Site VPN Connections section, check that the status of the one of the tunnels in Tunnel details is UP (since we made just one tunnel)
On AWS
Create an AWS t2.nano EC2 instance in your VPC. (Your VPC should already have a subnet else you will be asked to create one)
Let's call this instance AWS test host.
Verify that there is an internet gateway attached to your VPC.
Otherwise, choose Create Internet Gateway to create an internet gateway.
Select the internet gateway you created, and then choose Attach to VPC and follow the directions to attach it to your VPC.
In the navigation pane, choose Subnets, and then select your subnet.
On the Route Table tab, verify that there is a route with 0.0.0.0/0 as the destination and the internet gateway for your VPC as the target.
Otherwise, do the following:
Choose the ID of the route table (rtb-xxxxxxxx) to navigate to the route table.
On the Routes tab, choose Edit routes.
Choose Add route, use 0.0.0.0/0 as the destination and the internet gateway as the target.
Choose Save routes.
You should now be able to ssh into your instance.
If you still cannot ssh into your instance, check the instance inbound rules for SSH.
On Packet
Create a new server. We will call this Packet test host.
This means we have two machines from Packet. Previous was Packet VPN server and new one is Packet test host.
In this scenario, both machines are running Ubuntu 18.04.
We need to configure routing tables on both of them (static routes in this case since we are not using BGP).
ssh into Packet VPN server.
Add route to the tunnel IP on Packet side for requests going to the AWS VPC side.
# ip route add AWS_VPC_CIDR via PACKET_TUNNEL_IP
# e.g
ip route add 172.17.0.0/16 via 169.254.40.126
ssh into Packet test host.
Currently, Packet creates machines in separate subnets by default, so trying to fix routing tables here is somehow complicated. Packet test host can ping Packet VPN server but cannot reach AWS test host. In this case, we need to figure out how to tell the routing table of Packet test host that any request to anything in our AWS VPC should be routed through Packet VPN server. But how can this be done when both machines are on different subnets?
Packet.net's L3-only networking creates some challenges with routing packets from our packet.net private cluster to AWS. On packet.net, hosts are not in the same subnet. Instead, the host only sees a /31 network with an invisible datacenter-internal routing peer. In order to route packets to and from the VPN, we need to set up a tunnel between packet.net nodes and the VPN server. Those tunnels work peer-to-peer; i.e. while we need only one tunnel endpoint per node, the VPN server needs to have tunnel endpoints for all packet.net nodes. Please find an example configuration (server and one node) below. Please note that the interface names "node-1" (for Packet test host) and "vpn-server" are arbitrary and may be changed at will. We will use 192.168.0.0/30 for the internal tunnel transfer network, and we'll assume that the VPN server private IP is 10.80.166.1, and the node's IP is 10.80.166.3.
# VPN Server
vpnserver$ ip tunnel add node-1 mode ipip remote 10.80.166.3 local 10.80.166.1
vpnserver$ ip link set up dev node-1
vpnserver$ ip address add 192.168.0.1 dev node-1
vpnserver$ ip route add 192.168.0.0/30 dev node-1
vpnserver$ iptables -t nat -A POSTROUTING -s 192.168.0.2 -j SNAT --to 10.80.166.3
# Note that the last line (iptables) creates a source nat rule which hides the internal ipip tunnel transfer network from the AWS network.
# Node
node1$ ip tunnel add vpn-server mode ipip remote 10.80.166.1 local 10.80.166.3
node1$ ip link set up dev vpn-server
node1$ ip address add 192.168.0.2 dev vpn-server
node1$ ip route add 192.168.0.0/30 dev vpn-server
With this, you should be able to ping the AWS test host from Packet test host using their private IP addresses.
The end :)
Thank you @kosyfrances ! This helped me out with setting up some workflows.
Happy to mention also, that this works also for servers on a private LAN behind a NAT'd router. I set up the VPN tunnel and tested this with a KVM instance I created on my main PC behind my home router.
I had read that
NAT-Twas enabled automatically when using newer versions ofstrongswanand had also read that AWS had this available as well, so I decided to put it to the test and it worked.