I think at a minimum we have to have the behavior well-defined rather than open 
to interpretation. That IMO is the most important thing we got in IKEv2.

As long as some IKE SA exists between PEER-A and PEER-B, the peer that does not 
have the IPsec SA can inform the other side with an INVALID_SPI protected 
notification. But I agree it’s better to not get into this situation by wiping 
out the IPsec SAs with the IKE SA. Like we do in IKEv2.

> On 14 Jun 2018, at 19:03, riyaz talikoti <riyazin...@gmail.com> wrote:
> 
> Hi Yoav,
> Thanks a lot for the response.
> Just a quick question on top of what you wrote.
> 
> If a responder did not delete IPSec SA’s and only deletes IKE SA when it 
> receives IC.
> Will it not result in blackhole?
> 
> Lets say
> PEER-A ——————— — — PEER-B
> IKE SA(XX)                             IKE SA(XX)
> IPSEC SA(SPI AA)                   IPSEC SA(SPI AA)
> 
> PEER-A starts initiating new session and starts AGGRESSIVE_MODE.
> PEER-B receives IC and lets say PEER-B deletes ONLY IKE SA(XX)(recommended by 
> RFC),
> But will NOT delete IPSEC SA(SPI AA)(as RFC does not mention anything about 
> it).
> 
> And AM and QM exchanges gets completed and new sets of SA’s comes UP on both 
> sides
> 
> Now
> PEER-A ——————— — — PEER-B
> IKE SA(YY)                              IKE SA(YY)
> IPSEC SA(SPI BB)                   IPSEC SA(SPI BB)
>                                                 IPSEC SA(SPI AA) —> NOT 
> DELETED When IC is received.
> 
> For outgoing traffic from PEER-B, if SPI AA is chosen from SADB, will it NOT 
> get dropped at PEER-A with INVALID_SPI.
> So does it not make sense to delete all IPSec SA’s when IC is received.
> 
> Regards
> Riyaz
> 
> 
> 
>> On 12-Jun-2018, at 10:34 PM, Yoav Nir <ynir.i...@gmail.com 
>> <mailto:ynir.i...@gmail.com>> wrote:
>> 
>> 
>>> On 12 Jun 2018, at 11:53, riyaz talikoti <riyazin...@gmail.com 
>>> <mailto:riyazin...@gmail.com>> wrote:
>>> 
>>> Hi All,
>>> Need help with couple of questions related to INITIAL_CONTACT in IKEv1
>>> 
>>> 1. Is it NOT wrong to send INITIAL_CONTACT notification in QUICK MODE?
>>> Will it NOT end up in deleting the IKE SA(Phase 1 SA) which is being 
>>> created as part of just completed AGGRESSIVE MODE exchange?
>>> If we receive INITIAL_CONTACT notification in QUICK MODE, as a responder 
>>> should we ignore the notification?
>>> 
>>> 2. On receiving INITIAL_CONTACT we delete IKE SA. Doesn't it make sense to 
>>> delete all IPSec SA's(Phase 2 SA's) which are part of that particular IKE 
>>> SA(Phase 1 SA) ?
>>> Because the whole purpose is to inform responder to delete all previous 
>>> connection related to this identity as initiator is coming UP freshly.
>>> 
>> 
>> Hi, Riyaz
>> 
>> INITIAL_CONTACT is defined in section 4.6.3.3 or RFC 2407.  It does not 
>> specify that this notification should only be sent in phase 1 messages, 
>> although I agree that it makes little sense in the context of QUICK MODE.
>> 
>> So the answer to #1 is that it is not wrong.
>> 
>> The meaning of this notification is that all IKE SAs except the one being 
>> used or created in this negotiation is not known to the sender. RFC 2407 
>> unfortunately talks only of an “SA” without specifying whether this is about 
>> an IKE SA or an IPsec SA. I think the only sane interpretation is that this 
>> is about IKE SAs. So if an AGGRESSIVE negotiation created an IKE SA and this 
>> IKE SA is used to protect the QUICK MODE, that IKE SA is safe. If the 
>> AGGRESSIVE negotiation was used to create a different IKE SA, then it will 
>> be deleted.
>> 
>> You are always allowed to ignore an INITIAL_CONTACT notification. It is 
>> meant to help you with state management.  If you use some other IKE SA 
>> created before you received the INITIAL_CONTACT, you are very likely to get 
>> an error.
>> 
>> Regarding question #2: To clarify, on receiving INITIAL_CONTACT you delete 
>> all the other IKE SAs, not the one used in the negotiation.
>> The question of deleting the associated IPsec SAs when deleting an IKE SA 
>> is, unfortunately, not answered in the IKEv1 RFCs. Most vendors followed 
>> Cisco’s example and deleted them. A few didn’t.  In the case of 
>> INITIAL-CONTACT it makes even more sense than when an IKE SA is explicitly 
>> deleted with a DELETE payload.
>> 
>> HTH
>> 
>> Yoav
>> 
> 

Attachment: signature.asc
Description: Message signed with OpenPGP

_______________________________________________
IPsec mailing list
IPsec@ietf.org
https://www.ietf.org/mailman/listinfo/ipsec

Reply via email to