I want a Tunnel that use PFS: https://en.wikipedia.org/wiki/Forward_secrecy Client VPN is incomming, nerver use for outgoing connections, to i set responder only. For Outgoing i use Side 2 Side IKEv2. Mobike is a nice feature, so I turned on this crazy shit. At my point of view, Microsoft support it 7 and Apple up iOS 9 or newer this feature. https://tools.ietf.org/html/rfc4555

How I've set mine up, I've got a single phase 2 setup with the local network of 10.0.0.0/16 as I have VLANs with IP ranges 10.0.100.0/24, 10.0.200/24, 10.0.201.0/24, 10.0.202.0/24. I haven't got the "provide list of accessible networks" checked. I have allow rules for it to access my LAN subnets in the IPSec firewall rules. My client is windows and I can see that the 10.0.0.0/16 is showing up in my routing table as the second entry marked as on-link, all VLANs are accessible. Without seeing the routing table from your computer it's a bit hard to tell what it is trying to do.

If you are still having this issue, You could change your MODP (DH Group) on pfSense to 2 and the hash to sha1, that should get you past that point. Would be better to get the client side to use better settings though.

Thanks for that. I realise the pfSense version is badly out of date and needs updating. Unfortunately it has been planned then postponed by the people that own the firewalls several times after I'd done all of the work to prepare. It was discussed again a few days ago and the hardware will be replaced with a fresh pfSense install. I believe they had CARP break in an upgrade between 2.1.x to 2.2 and have been reluctant to upgrade them ever since.

IPsec policies match before the routing table. Policy routing happens before IPsec policies and the routing table. But only when the connection is initially established into that interface. Glad you found it.

piggy back off this thread, I have a mobile client ipsec tunnel set up on site a, and I've been trying to figure out a way for that mobile client (sub net 192.168.117.x) to reach site b in that hub and spokes structure. I tried adding a new p2 to the site a -> hub p1 with local subnet being 192.168.117.x remote subnet being the hub subnet. On the hub I added p2 with local being hub subnet and remote 192.168.117.x subnet. So far the mobile client can't connect to hub. I'm not too experienced with setting up tunnels using ipsec, does anyone have success in setting up similiar network?

Well this is beginning to look like a weird IPsec bug to me! Home Office Phase 2 Entries: 192.168.10.0/24 to 192.168.40.0/24 ; Works! 192.168.11.0/24 to 192.168.40.0/24 ; Works! 10.1.12.0/24 to 192.168.40.0/24 ; Does Not Work Remote Office Phase 2 Entries: 192.168.40.0/24 to 192.168.10.0/24 ; Works 192.168.40.0/24 to 192.168.11.0/24 ; Works 192.168.40.0/24 to 10.1.12.0/24 ; Does Not Work Update: The problem seems to be isolated to the remote office firewall as it has the same problem with a peer-to-peer OpenVPN tunnel and other remote offices work fine with both IPsec and OpenVPN tunnels in the same configuration.

I recently got hit with this, couldn't run remote backups of my switches via SSH over a VTI tunnel after switching from site to site. I noticed that entering 1400 in both the MTU and MSS for the interface resulted in a scrub for max-mss to 1360 which seems to be correct. Not sure why pfsense auto deducts 40 from the MSS, since MSS should be 40 less than MTU already. Using 1400 in both fields on both ends of the links has resolved my issues here.

@jgraham5481 I did not explain myself properly with my writing. I can see how it could read that way. The domain controller has a NPC and also a DHCP and DNS services running. NPC provides the Radius server and the policies for authentication and authorization. pfsense is managing routing between the WAN and LAN and it is assigned as the Gateway by the DHCP server (hosted in the DC) for all the devices in the LAN that use dynamic IP. When a client joins the VPN, it does in a different subnetwork with IP range assigned by pfsense. The latter does the routing between the VPN subnet and the LAN subnet for the VPN clients.

@SuperHans If this is for mobile, you pretty much have to choose any, unless you have something it can match. If using like radius and nps, it can’t determine user distinguished name by default. By setting to any, it just means all of the other auth requirements must be met, ie: cert, username and password, etc.

@hipgroove What do your proposals look like? There’s a sweet spot for mac and a sweet spot for windows 10.

@marcquark Hi I also struggled with SA duplicates for a long time , and as a result I wrote my own small program ( С ) that interacts with FreeBSD kernel ( PF_KEY Key Management API ) and Strongswan ( Vici and Stroke Protocol ) . This program receives messages from the kernel when a new SA is installed and checks how many are currently in the system . If it finds a duplicate, it sends a Strongswan message to delete the duplicate Here's what it looks like Jun 16 10:05:43 daemon sa_checker started Jun 16 10:05:43 [KNL] successfully connected to PF_KEY_V2 socket Jun 16 10:05:54 [KNL] received message from kernel,type of message SADB_DELETE,errno = 0 ,seq number = 53850 Jun 16 10:05:54 [KNL] parsed data : SPI = 0xcecde75d , SRC ADDRESS = 10.3.100.100 , DST ADDRESS = 10.3.100.1 Jun 16 10:05:54 [KNL] received message from kernel,type of message SADB_ADD,errno = 0 ,seq number = 53851 Jun 16 10:05:54 [KNL] parsed data : REQID = 0x3e8, SPI = 0xcecde75d , SRC ADDRESS = 10.3.100.100 , DST ADDRESS = 10.3.100.1, SA TYPE = IPsec ESP Jun 16 10:05:54 [KNL] received message from kernel,type of message SADB_ADD,errno = 0 ,seq number = 53852 Jun 16 10:05:54 [KNL] parsed data : REQID = 0x3e8, SPI = 0xc459c817 , SRC ADDRESS = 10.3.100.1 , DST ADDRESS = 10.3.100.100, SA TYPE = IPsec ESP Jun 16 10:05:55 [KNL] successfully connected to stroke socket Jun 16 10:05:55 [KNL] stroke message successfully sended,waiting for reply .... Jun 16 10:05:55 [KNL] server reply is ' con1000{243}: INSTALLED, TUNNEL, reqid 1000, ESP SPIs: cb0c99c0_i c0d72295_o ' Jun 16 10:05:55 [KNL] server reply is ' con1000{244}: INSTALLED, TUNNEL, reqid 1000, ESP SPIs: cecde75d_i c459c817_o ' Jun 16 10:05:55 [IKE] total child's SA for deleting 1 Jun 16 10:05:55 [IKE] CHILD SA for deleting 243 Jun 16 10:05:55 [IKE] closing CHILD_SA con1000{243} with SPIs cb0c99c0_i (172 bytes) c0d72295_o (71116 bytes) and TS 0.0.0.0/0|/0 === 0.0.0.0/0|/0 Jun 16 10:05:55 [IKE] sending DELETE for ESP CHILD_SA with SPI cb0c99c0 Jun 16 10:05:55 [ENC] generating INFORMATIONAL request 1535 [ D ] Jun 16 10:05:55 [NET] sending packet: from 10.3.100.1[500] to 10.3.100.100[500] (80 bytes) Jun 16 10:05:55 [NET] received packet: from 10.3.100.100[500] to 10.3.100.1[500] (80 bytes) Jun 16 10:05:55 [ENC] parsed INFORMATIONAL response 1535 [ D ] Jun 16 10:05:55 [IKE] received DELETE for ESP CHILD_SA with SPI c0d72295 Jun 16 10:05:55 [IKE] CHILD_SA closed Jun 16 10:05:55 [IKE] CHILD SA with uniqueid 243 has deleted successfully Jun 16 10:05:55 [KNL] received message from kernel,type of message SADB_DELETE,errno = 0 ,seq number = 53853 Jun 16 10:05:55 [KNL] parsed data : SPI = 0xc5cdc8dd , SRC ADDRESS = 10.3.100.102 , DST ADDRESS = 10.3.100.1 Jun 16 10:05:55 [KNL] received message from kernel,type of message SADB_ADD,errno = 0 ,seq number = 53854 Jun 16 10:05:55 [KNL] parsed data : REQID = 0x7d0, SPI = 0xc5cdc8dd , SRC ADDRESS = 10.3.100.102 , DST ADDRESS = 10.3.100.1, SA TYPE = IPsec ESP Jun 16 10:05:55 [KNL] received message from kernel,type of message SADB_ADD,errno = 0 ,seq number = 53855 Jun 16 10:05:55 [KNL] parsed data : REQID = 0x7d0, SPI = 0xcd2b03b8 , SRC ADDRESS = 10.3.100.1 , DST ADDRESS = 10.3.100.102, SA TYPE = IPsec ESP Jun 16 10:05:56 [KNL] successfully connected to stroke socket Jun 16 10:05:56 [KNL] stroke message successfully sended,waiting for reply .... Jun 16 10:05:56 [KNL] server reply is ' con2000{242}: INSTALLED, TUNNEL, reqid 2000, ESP SPIs: ca05bb45_i c4a73853_o ' Jun 16 10:05:56 [KNL] server reply is ' con2000{245}: INSTALLED, TUNNEL, reqid 2000, ESP SPIs: c5cdc8dd_i cd2b03b8_o ' Jun 16 10:05:56 [IKE] total child's SA for deleting 1 Jun 16 10:05:56 [IKE] CHILD SA for deleting 242 Jun 16 10:05:56 [IKE] closing CHILD_SA con2000{242} with SPIs ca05bb45_i (0 bytes) c4a73853_o (216292 bytes) and TS 0.0.0.0/0|/0 === 0.0.0.0/0|/0 Jun 16 10:05:56 [IKE] sending DELETE for ESP CHILD_SA with SPI ca05bb45 Jun 16 10:05:56 [ENC] generating INFORMATIONAL request 127 [ D ] Jun 16 10:05:56 [NET] sending packet: from 10.3.100.1[500] to 10.3.100.102[500] (80 bytes) Jun 16 10:05:56 [NET] received packet: from 10.3.100.102[500] to 10.3.100.1[500] (80 bytes) Jun 16 10:05:56 [ENC] parsed INFORMATIONAL response 127 [ D ] Jun 16 10:05:56 [IKE] received DELETE for ESP CHILD_SA with SPI c4a73853 Jun 16 10:05:56 [IKE] CHILD_SA closed Jun 16 10:05:56 [IKE] CHILD SA with uniqueid 242 has deleted successfully If you are interested, I can give you a program for the test

@tschmit I have same problem, client OSX 10.13 native client. All work fine with 2.4.4p3 but is broken with 2.4.5p1

I don't see default routes like /0 networks in my IPSec policy. Did you set it up for mobile clients or site-to-site? manual config of IPSec.conf won't cut it. I think the entire ipsec directory is generated on demand. Can you should your full ipsec.conf

@jimp for the time being, it'd be nice to have the first 64 bits tracking one of the LAN interfaces and the last 64 a user-definable range. I think this is possible through some grepping while generating ipsec.conf with rightsourceip = <globally routable prefix> + <user-defined range> I want to give clients IPv6 internet access and avoid NAT or manually defining the first 64 bits

@jimp I verified it. I change it to a RSA 2048 cert, it successfully loaded cert and I can establish conn

Dear all, with the help of https://docs.netgate.com/pfsense/en/latest/vpn/ipsec/configuring-an-ipsec-remote-access-mobile-vpn-using-ikev2-with-eap-mschapv2.html I managed to configure a IKEv2 SA and child SA for the ESP IPsec tunnel for my iPhone iOS v13.5.1 to pfSense 2.4.5. I created a new PKI and converted the client certificate .p12 with a OpenSSL lib workaround I found here. And after trying a while it works now for me. IPsec connection establishes fast and reliable. But what I still do not understand with the above method is: Why do I need to define PSK keys for the EAP authentication part after IKE handshake although I already have a client certificate in place on the mobile that actually could also do this job (or even better). iPhone allows to configure IKE connections to use the certificate as user authentication method. But with this method set (instead of the user / pw pattern) I cannot manage to authenticate successfully. EAP authentication of the client (iPhone) always gets aborted: Last 1000 IPsec Log Entries. (Maximum 1000) 09[IKE] <con-mobile|2> IKE_SA con-mobile[2] state change: CONNECTING => DESTROYING 09[NET] <con-mobile|2> sending packet: from <ServerIP> [4500] to <iPhoneIP>[19330] (80 bytes) 09[ENC] <con-mobile|2> generating IKE_AUTH response 3 [ EAP/FAIL ] 09[IKE] <con-mobile|2> received EAP_NAK, sending EAP_FAILURE 09[ENC] <con-mobile|2> parsed IKE_AUTH request 3 [ EAP/RES/NAK ] 09[NET] <con-mobile|2> received packet: from <iPhoneIP>[19330] to <ServerIP>[4500] (80 bytes) 09[NET] <con-mobile|2> sending packet: from <ServerIP>[4500] to <iPhoneIP> [19330] (112 bytes) 09[ENC] <con-mobile|2> generating IKE_AUTH response 2 [ EAP/REQ/MSCHAPV2 ] 09[IKE] <con-mobile|2> initiating EAP_MSCHAPV2 method (id 0xDC) 09[IKE] <con-mobile|2> received EAP identity 'Markus' 09[ASN] <con-mobile|2> file content is not binary ASN.1 09[ENC] <con-mobile|2> parsed IKE_AUTH request 2 [ EAP/RES/ID ] ... Do you have an idea? kind regards Markus