7

I have an issue with a mixture of remote access, L2L and dynamic L2L tunnels on an ASA5540 running 8.2

Here is a snip of the relevant config:-

crypto dynamic-map outside-crypto-dynamic-map 10 match address outside-crypto-dynamic-map-10
crypto dynamic-map outside-crypto-dynamic-map 10 set transform-set ESP-3DES-MD5
crypto dynamic-map outside-crypto-dynamic-map 20 set transform-set ESP-3DES-MD5
crypto map outside-crypto-map 201 match address outside-crypto-map-201
crypto map outside-crypto-map 201 set peer X.X.X.X
crypto map outside-crypto-map 201 set transform-set ESP-3DES-MD5
crypto map outside-crypto-map 202 match address outside-crypto-map-202
crypto map outside-crypto-map 202 set peer Y.Y.Y.Y
crypto map outside-crypto-map 202 set transform-set ESP-AES256-SHA
crypto map outside-crypto-map 65535 ipsec-isakmp dynamic outside-crypto-dynamic-map
crypto map outside-crypto-map interface outside

I have a number of remote sites that use dynamic IPs. The LAN subnets for these are in an ACL "outside-crypto-dynamic-map-10".

These match fine based on this line:-

crypto dynamic-map outside-crypto-dynamic-map 10 match address outside-crypto-dynamic-map-10

I have other "static" L2L tunnels that work fine per 201 and 202 in the config above.

With my remote access (Cisco VPN Client) users, unless I have the following line, they won't connect:-

crypto dynamic-map outside-crypto-dynamic-map 20 set transform-set ESP-3DES-MD5

If I try to add a "match address" statement to that sequence (like below), then remote access stops working (where "vpc-client-subnet" is an ACL containing the subnet from the IP pool used for the remote access clients) as it cannot find a matching local/remote.

crypto dynamic-map outside-crypto-dynamic-map 20 match address vpc-client-subnet

The problem is that I have a number of L2L endpoints out there (that I no longer want connected but have no control over) that are still configured (on the remote end) with the PSK used by the dynamic L2L peers. I have removed their LAN subnets from "outside-crypto-dynamic-map-10" (as I don't want them connected any more) so they don't match on the dynamic-map seq 10 anymore, however they are still able to successfully complete phase 2 against "outside-crypto-dynamic-map 20" and it seems as if this end just accepts whatever the remote proposes as the local/remote for the SA.

I am not in a position to change the PSK. I can't add a "match address" to the "outside-crypto-dynamic-map 20" as it will prevent remote access clients from connecting, however if I don't, then it acts as a catch-all for other peers that otherwise know the PSK.

Ideally I could add "match address" to seq 20, so that remote access users will match against it, but the "old" seq 10 peers will not. Alternatively, is there a way to prevent the ASA from accepting the remote peers idea of the local/remote for the SA when it has matched against a map that doesn't have a "match address" statement?

EDIT:

The relevant tunnel group config:-

tunnel-group DefaultL2LGroup ipsec-attributes
 pre-shared-key *****
 peer-id-validate nocheck
 isakmp keepalive threshold 30 retry 5

tunnel-group x.x.x.x type ipsec-l2l
tunnel-group x.x.x.x ipsec-attributes
 pre-shared-key *****
 isakmp keepalive threshold 30 retry 5

tunnel-group ravpn type remote-access
tunnel-group ravpn general-attributes
 address-pool ip-pool-ravpn
 authentication-server-group Edirectory
 default-group-policy ravpn
tunnel-group ravpn ipsec-attributes
 pre-shared-key *****
 isakmp keepalive threshold 30 retry 2

tunnel-group-map default-group DefaultL2LGroup

There are multiple tunnel-groups for the "static" tunnels (per the x.x.x.x example above). The remote access users hit the "ravpn" tunnel-group, and the dynamic ones all match the "DefaultL2LGroup". The PSK is not the same between these.

6
  • Hi, Can you post show run tunnel-group? Are you using the same PSK both for RA clients and dynamic IPSEC peers?
    – laf
    Aug 13, 2013 at 11:55
  • @laf I've updated the question to include the information you requested. Thanks.
    – aleks
    Aug 15, 2013 at 6:20
  • Hi, I was on vacation, so I apologize for the delay; hope it will be us though. What I wanted to check is whether those unwanted connections have each a tunnel group defined or if all of them are using the same PSK from ravpn group. What is the status of this?
    – laf
    Aug 21, 2013 at 13:35
  • Any news on this?
    – laf
    Aug 26, 2013 at 9:44
  • @laf the unwanted connections are using the PSK from the DefaultL2LGroup; they do not have their own tunnel groups
    – aleks
    Aug 28, 2013 at 0:14

3 Answers 3

2

I would suggest to create a 2nd Remote-Access tunnel group for all your remote-access users. Start migrating them, one by one and when you are done you can delete the existing Default L2L Group, or you can change the PSK.

If you encounter some users that don't want to change, you can enforce a specific VPN Client version and inform them you are doing a major security update.

0

Typically your remote VPN users only need a couple lines under the dynamic crypto map that allow them to connect properly. Something like this:

crypto dynamic-map DYN_CRY_MAP 65535 set transform-set ESP-AES-128-SHA

crypto map OUTSIDE_map 65535 ipsec-isakmp dynamic DYN_CRY_MAP

This is assuming you have the appropriate matching transform-set already specified.

Your remote user VPN subnets do not need to be specified in the crypto map since they are dynamic endpoints. The ASA knows this and adjusts the the crypto tables accordingly. Your dynamic L2L endpoints also do not need to be specified under the crypto map and will operate in the same way. Also, a "feature" is that the last of the two lines above MUST be applied at the very end of your crypto map. You already have it there, but just wanted to let you know.

This line is what allows Phase 1 to connect from a dynamic L2L endpoint:

tunnel-group DefaultL2LGroup ipsec-attributes

 pre-shared-key *****

 peer-id-validate nocheck

 isakmp keepalive threshold 30 retry 5

Phase 2 is handled by your crypto map entries.

I'm afraid that the easiest and most effective way to do this would be to change the PSK on the DefaultL2LGroup tunnel-group and then change it on all of the devices that you still want connected. Since they're dynamic endpoints, you can't really filter by IP on the outside interface, and you have no control over them establishing a tunnel if they know the PSK.

If you have any questions, please don't hesitate to ask.

Ian

0

@aleks - If you know which L2L devices you no longer want to be allowed, and you know their private IP ranges, couldn't you just deny those private IPs on your Inbound ACL (the one applied to your ASA's outside interface). This might not prevent Phase 2 setup, but it would for sure not allow the traffic to pass. Just a thought (I ran across this question while googling for a different issue).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.