When trying to create a VPN tunnel between a CheckPoint firewall and Amazon managed VPN service I happened upon a unpleasant surprise.
Namely when using stronger crypto methods than defined by default in the guides by CheckPoint or Amazon you will run in to an issue, that the CheckPoint device will start dropping traffic after Phase2 key exchanges for a ~5 minute time period. To be more exact the traffic from Amazon to the hosts/networks behind the CheckPoint GW will start failing and connections started from behind the CheckPoint device will continue working as before. Namely Amazon VPN service refreshes it’s keys 5 minutes before the lifetime set in the VPN properties and CheckPoint close to 30 seconds. It actually wouldn’t be a problem if Amazon would use the same parameters as were used to initially establish the tunnel, but it doesn’t. It will actually use DH group 2 to initiate key exchange after which the CheckPoint device will start dropping the traffic coming in from the Amazon service with the following error:
encryption failure: Packet was decrypted with methods which are different from the methods according to the security policy - Gateway and Peer use different DH groups
After talking to both CheckPoint and Amazon support, I can say that the only thing you can do to remedy this is actually setting the DH group to 2 for PFS.
Although Amazon in its documentation(here) states it supports a bunch of different DH groups, and yet it defaults to DH group 2 when initiating the connection it self. To be honest, to me it seems a bit strange that the AWS VPN actually mirrors the encryption/integrity settings of the previous negotiation, but doesn’t remember the PFS settings and defaults to DH group 2. When talking to support services the only thing that AWS support suggested was to force the CheckPoint device to exchange keys before the AWS service does. Unfortunately you cannot do that according to Check Point support services, as there is no such setting available and that timer is around 30s+- some random number of seconds prior to the end of the life time set in the VPN properties.
May 25, 2018 at 10:41
Thanks a lot for the tips, we are having the exactly same problem here between CP and AWS.
I was having the same feeling that AWS was dropping the dh config to default or not accepting the dh14, in my case connection was dropping every hours for couple of minutes.
I make the change now to DH2 aswell, will keep monitoring 😉
Thanks alot for your help, sharing is caring 😀