B2B VPN IKEv2 Fail with Amazon Private Cloud Peer

[pnelson]

Setting up a VPN with a vendor. It came up the first time and test data was passed. It was a couple of weeks after testing before the tunnel would actually be used. When that time came we could not get the tunnel up. IKEv2 fails. I know nothing changed on my end and the vendor makes the same claim. The vendor is using Amazon virtual private cloud firewall. I am using an on-prem PA-3250 HA pair (active/passive) inside of my Internet perimeter with 96 perfectly happy B2B VPNs running. We have tried everything to get the tunnel back up including tearing down the config on both peers and rebuilding from scratch, my side using a new tunnel interface.

The vendor has to initiate the tunnel as the tunneled host on my side only sends stateful replies, the vendor always sends the SYN, so my side is always the IKEv2 responder.

The IKEv2 initiator request comes in and my side responds. SPIs are established so we have peer-to-peer communication. No issues with NAT-T.

I have attached a packet capture of an IKEv2 exchange (4 packets) and a system log of the failure (2 lines). Digging into the packets I see key exchange and it looks like the Phase 1 proposal is correct. Would someone take a look and educate me on what I'm missing? Thanks.

Pete

[pnelson]

I think I have found the issue.  The initiating peer has an incorrect peer ID for my side (responder) as revealed by CLI 'less mp-log ikemgr.log'.  It gives more detail than the GUI system log.  My vendor counterpart is on the other side of the globe so I'm not sure when he will be attempting the correction.

[pnelson · Accepted solution]

This issue has been resolved.  On my side I had to set the local peer ID to my firewall interface's public NAT address, not the native address.  IKEv2 still wouldn't come up so the vendor changed his proposal from being specific to throwing many combinations my way.  After that Phase 1 & Phase 2 came up instantly.