
IPsec VPN troubleshooting starts with the traffic that fails. A site-to-site tunnel can show as up while applications still fail. Start with one known source, destination and application port. Compare IPsec packet counters at both gateways during that test to see whether packets enter the tunnel, reach the remote site and return.
1. Verify the route in both directions
Check the route to the remote host from the local gateway and the return route from the remote gateway. In a route-based VPN, verify that the remote subnet resolves through the intended tunnel interface in the correct routing table or VRF. In a policy-based VPN, check that the flow reaches the interface where the VPN policy applies. A more specific route or a policy route can send the test traffic elsewhere.
2. Check traffic selectors and NAT
For policy-based VPNs, compare the negotiated local and remote selectors on both peers. Each gateway’s local subnet should match the other gateway’s remote subnet for the intended flow. Some route-based tunnels negotiate broad selectors. In that case, routing and firewall policy still determine which traffic can pass.
Check the effective NAT rule for the test flow. If ordinary internet source NAT changes the address before VPN policy evaluates it, the resulting packet may no longer match the intended selectors. Verify the rule order and use the translated addresses when the VPN was explicitly designed to carry them. For background, see how IPsec VPN works.
3. Use packet counters for IPsec VPN troubleshooting
On Cisco IOS and ASA, show crypto ipsec sa displays encapsulation and decapsulation counters. Select the SA for the correct peer and traffic selectors. Record counters before and after a small test from an actual LAN host. Busy tunnels may need flow logs to distinguish your test from unrelated traffic.
show crypto ipsec saLocal encaps stays flat. Check the source host’s gateway, the selected route and whether the flow matches VPN policy after NAT.
Local encaps rises but remote decaps stays flat. Check the path between gateways and whether the receiving peer sees or rejects the encrypted packets. Review the expected encapsulation for your design, including UDP 4500 when NAT traversal is in use.
Remote decaps rises but no reply returns. Check the remote firewall policy, server listener, host firewall and return route. Then check whether the remote encaps counter rises when the server responds.
Worked example: Requests reach Site B but replies never leave
In this illustrative example, a client at Site A tries to open an internal HTTPS service at Site B. Site A encaps and Site B decaps both increase, while Site B encaps stays flat. The request crossed the tunnel. Check Site B’s gateway logs and the server-side path, then confirm that the server replies and that its return route points toward the Site B VPN gateway.
Verify the application after the fix
Repeat the original application test from the original LAN host. Confirm that counters rise in both directions on the expected SAs. Check the application response as well as the network path. If small probes work but larger transfers stall, test MTU and fragmentation separately. See Cisco’s IKEv2 site-to-site verification guide for platform-specific counter examples.
Cover photo by Brett Sayles via Pexels.

0 Comments