Skip to main content

IPv6 IPsec VPN Tunnel Palo Alto <-> FortiGate

VPN tunnels will be used over IPv6, too. I configured a static IPsec site-to-site VPN between a Palo Alto Networks and a Fortinet FortiGate firewall via IPv6 only. I am using it for tunneling both Internet Protocols: IPv6 and legacy IP.
While it was quite easy to bring the tunnel “up”, I had some problems tunneling both Internet Protocols over the single phase 2 session. The reason was some kind of differences within the IPsec tunnel handling between those two firewall vendors. Here are the details along with more than 20 screenshots and some CLI listings.

(Please note that I have many different VPN tutorials on my blog. Have a look at this list to find the appropriate post. This one here focusses on IPv6 tunneling.)

Lab

My lab consists of a Palo Alto Networks PA-200 firewall with PAN-OS 8.0.3, and a Fortinet FortiWiFi 90D with Firmware Version v5.4.5, build1138. I am using some uncommon but highly secure crypto protocols: Diffie-Hellman group 20 (have a look here), AES-256, SHA-512 and a lifetime of 28800 s (IKE) respectively 3600 s (IPsec). The following figure shows the IP addressing scheme. Note that the VPN tunnel is established over IPv6 only while it tunnels IPv6 and legacy IP!

The configuration was almost straightforward. However, it took me a while to understand the handling of the phase 2 sessions: While Palo Alto simply establishes a single phase 2 tunnel and forwards IPv6 as well as IPv4 packets through it, FortiGate needs two different phase 2 tunnels, one for IPv6 and one for IPv4. That is: I configured two Proxy IDs on the Palo as well, one for IPv6 and another for IPv4. Here are some information from Forti that helped me in thinking about several phase 2 settings.

Configuration Palo Alto

The configuration of the Palo firewall consists of the following steps: IKE Gateway, Tunnel Interface, IPsec Tunnel with Proxy IDs for IPv6 and IPv4, static routes for IPv6 and IPv4, dual-stack policies. Here we go:

Configuration FortiGate

Except the tunnel interface (which must not be added separately) and two separate policy sets (since FortiGate has a shit policy design which distinguishes between the Internet Protocols) the config on the FortiGate is very similar: IPsec Tunnel with Gateway, Authentication, Phase 1 Proposal and two Phase 2 Selectors (IPv6 and IPv4), as well as two static routes (IPv6 and IPv4) and four policies (IPv6 and IPv4). Let’s do this:

Monitoring

I had two Ubuntu clients, one behind each firewall. Rather than only pinging I did some file transfers via ssh/scp. Here are some traffic logs from both firewalls:
At the Tunnel Info from Palo Alto you can see the odd behaviour due to the phase 2 tunnel handling. All outgoing packets are sent via the IPv4 tunnel “fg:fg”. The counter for “Pkt Encap” at the IPv6 tunnel “fg:fg6” shows a 0:

On the FortiGate everything seems to be ok. The counters increased for both phase 2 tunnels, i.e., IPv6 and legacy IP.

Here are some CLI outputs from the Palo Alto:
And some CLI outputs from the FortiGate as well:

Conclusion

Yeah it works. Great. Two different firewall vendors, IPv6, uncommon protocols (DH20, SHA512), different Proxy ID handling, but still: it works. Keep on going, guys!
(But I am still a bit irritated about the phase 2 tunnels. Which is better? Palo Alto: simply establishing a single phase 2 and handling everything over it or FortiGate: different phase 2 tunnels for each Internet Protocol? I don’t know. But good to know how to circumvent it.)

Comments

Post a Comment

Popular posts from this blog

Checkpoint firewall common commands part 2

Checkpoint firewall common commands part 2 For basic firewall informaton gathering: fgate stat -Status and statistics of Flood-Gate-1. fwaccel <stat|stats|conns>  – View status, statistics or connection table of SecureXL. fw getifs -Show list of configured interfaces with IP and netmask. cpstat <app_flag> [-f flavour] -View OS, HW and CP application status. Issue cpstat without any options to see all possible application flags <app_flag> and corresponding flavours. Examples: cpstat fw -f policy – verbose policy info cpstat os -f cpu – CPU utilization statistics cpinfo -y all   -List all installed patches and hotfixes. cpd_sched_config print -Show task scheduled with CPD scheduler. enabled_blades -View enabled software blades avsu_client [-app <app>]   , get_version <app>  -Get signature version and status of content security .Without the -app option “Anti Virus” is used. show co...
Post a Comment
Read more

Unable to Connect to Server Checkpoint R80

Unable to Connect to Server Checkpoint R80 Unable to Connect to Server A connection to the management server will fail if: A firewall between SmartConsole and the management server blocks Port 19009 -  port 19009 is used for a new R80 service. Allow traffic on this port for all clients and management servers. No GUI clients are assigned -  Open the Gaia Portal. If the First Time Configuration Wizard opens, complete it. If the First Time Configuration Wizard has already run, open  User Management > GUI Clients  and add a client. When using Multi-Domain Security Management, connect SmartConsole to the Multi-Domain Server and make sure the domains have GUI clients assigned to them. The required processes are not reachable -  Make sure the computer with SmartConsole installed can reach the IP address of the management server, and that these server processes are up and running: cpm fwm Operation time out  – Your connection ...
Post a Comment
Read more

Configuring Proxy ARP for Manual NAT

Configuring Proxy ARP for Manual NAT Symptoms After creating a Manual Static NAT rule, Security Gateway does not answer the ARP Requests for the Static NATed IP address that was configured in the Manual NAT rule. Security Gateway replies to ARP requests with a wrong MAC address, mostly for the NAT traffic.  Introduction Let us consider the following scenario: Two networks ( Network_A  and  Network_B ) are separated by a Security Gateway (single Security Gateway or ClusterXL). On each network, there is a host ( Host_A  on  Network_A ,  Host_B  on  Network_B ). Let us assume, that  Network_A  represents the  Internal  network, and  Network_B  represents the  External  network. According to the existing standards, when  Host_B  needs to send data to  Host_A , an ARP Request for the MAC address of  Host_A  will be sent by  Host_B  to  Network...
Post a Comment
Read more