Re: [I2nsf] [Last-Call] New Version Notification for draft-ietf-i2nsf-sdn-ipsec-flow-protection-11.txt

[tom petch <[e-mail]>]

Rafa

As you may know, the submission window closes in the week before the 
IETF meeting which is perhaps why over 150 new I-D have just appeared:-( 
which may or may not affect your plans. I expect to look some more this 
week but the country is going into lockdown for four weeks, some are 
saying for six months, so I need to work out what I have to get done 
before all the places I need to go shut:-(  Some things pre-empt IETF work!

On the formatting, my hope was that the paragraphs of text as in YANG 
description would spread out to occupy fewer lines, so fixing the 
indentation is an improvement but not quite all I had hoped for - it 
certainly makes it easier to read.  I would not reissue a new version 
just for this

And inline twice


On 30/10/2020 17:43, Rafa Marin-Lopez wrote:
> Hi Tom:
>
> Thank you very much again. See our comments inline:
>
>> El 29 oct 2020, a las 18:06, tom petch <[e-mail]> escribió:
>>
>> Back to routine comments on this I-D; I don't think that any of my earlier ones are unresolved.
>>
>> On 29/10/2020 11:23, Rafa Marin-Lopez wrote:
>>> Hi Tom:
>>>
>>>> El 28 oct 2020, a las 19:02, tom petch <[e-mail]> escribió:
>>>>
>>>> On 28/10/2020 10:42, Rafa Marin-Lopez wrote:
>>>>> Hi Tom:
>>>>>
>>>>> Thank you very much for your insight. It is very helpful. Please see our comments/questions inline.
>>>>>
>>>>>> El 27 oct 2020, a las 13:42, tom petch <[e-mail]> escribió:
>>>>>>
>>>>>> I think that the IESG will find a number of problems with this I-D.
>>>>>>
>>>>>> YANG module references RFC822 which is several years out of date
>>
>> Rafa
>>
>> Several YANG leaf are in seconds, eight I think, but none use the YANG
>> units seconds;
>> statement.  I suggest adding it.
>
> Added. They are seven.

I still see eight - the one that lacks units is
leaf established

>> typedef ipsec-spd-action
>> capitalises the actions as does RFC4301 but later uses of this have lower case actions; probably ok as it is
>
> Ok.
>>
>> protocol number appears in several places but technically, IPv6, which we must all love and promote, has 'next header' instead.  Probably worth a mention at least once in the YANG module
>
> We have included this note:
>
> NOTE: In case of IPv6, the protocol in the IP packet payload is
> specified in the Next Header field of the IPv6
> packet.";
>>
>> On the start >= end, I was looking at nat-yang and it has
>> must ". >= ../start-port-number"
>> which is the sort of YANG constraint I had in mind
>
> Done.
>
> must '. >= ../start’
>
>> leaf dscp-mapping
>> I do not understand, a hex string of indeterminate length for DSCP values (plural).  I expect DSCP to be six bit and values to be a leaf-list
>
> Right. Since we have inet:dscp type , we have used that one in a leaf-list as you recommend.
>
> leaf-list dscp-mapping {
>         type inet:dscp;
>         default 0;
>         description
>           "DSCP values allowed for packets carried over
>           this IPsec SA.";
>         reference
>           "Section 4.4.2.1. in RFC 4301.";
>       }
>
>>
>> leaf ecn
>> /Annex C/Appendix C/
>
> Fixed.
>>
>> I see that RFC8247 gives AEAD status which is a good reference to have except that as my other post today says there is no requirement for any future additions to the IANA data to include this information; IANA for IKEv2 needs updating iun a separate I-D!
>>
>> leaf mask
>> hex-string with no length specified
>
> Since this masked is after all 32-bits mask we have used uint32 instead.
>>
>> leaf ds-algorithm
>> default 1
>> which I think is RSA; worth stating as it might raise some AD eyebrows
>
> Clarified in the description.
>>
>> pfs-groups
>> /it is required perfect forward secrecy/perfect forward secrecy is required/
>
> Fixed.
>
>> and a reference to Transform Type 4?
>
> As we mention earlier, this pfs-groups used the type pfs-group, which includes the reference to Transform Type 4.
>>
>> leaf ext-seq-num
>> /FALSE/false/ or /False/?
>>
>> leaf encryption-algorithm (twice, in separate containers)
>
> Mmmmm… we only see in one place.

I see it a second time for
leaf integrity-algorithm


Tom Petch

>> /the integrity leaf/ the integrity-algorithm leaf/
>
> Fixed.
>>
>> I have yet to look at the text or the examples, probably next week.
>
> Thank you very much. We are going to post -12 with all these comments and the previous ones so far. If there are pending comments we can prepare -13 right away.
>
> Best Regards.
>>
>> Tom Petch
>>
>>>> On boiler plate, I mean the reference to RFC2119 which now must use the language from RFC8174 in the body of the I-D; sorry for the confusion.
>>>
>>> Ah ok. I guess you are referring to change that paragraph with this:
>>>
>>>
>>> The key words "MUST", "MUST NOT", "REQUIRED",
>>> "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT",
>>> "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
>>> "OPTIONAL" in this document are to be interpreted
>>> as described in BCP 14 RFC2119 RFC8174 when,
>>> and only when, they appear in all capitals,
>>> as shown here.
>>>
>>>>
>>>> On XXXX, you have XXXX standing in for more than one RFC-to-be which confuses.  The convention is to use XXXX for this I-D and then AAAA BBBB etc for any others such as the netconf I-D in this instance.
>>>
>>> Ah ok. In any case, we have now only XXXX for our RFC-to-be so problem fixed.
>>>>
>>>> Two big issues, for me (perhaps not for others).  The convention with YANG is for each successive line to be indented two characters, you have four, which creates a lot of white space and pushes the text to the right hand margin.  I think that two characters is the default when you use pyang to format a YANG module.
>>>
>>> We can try to reduce it to two characters.
>>>
>>>>
>>>> And references.  I have had to work harder than I want to to make sense of the IANA references.  I think you should have five separate references in the I-D for IANA for
>>>> Transform Type 1
>>>> Transform Type 3
>>>> Transform Type 4
>>>> Authentication Method
>>>> Protocol Numbers
>>>> and each reference in the I-D should have a URL pointing to the specific section of IANA web site.
>>>
>>> Ok, good. This follows what we thought.
>>>
>>>> In the YANG, it is harder to know what to do.  Those first three references are in the third tier i.e.
>>>> Group - Internet Key Exchange V2 (IKEv2) Parameters
>>>> Registry  -Transform Attribute Types
>>>> and then Type 1, 3, 4 as the third tier as I am calling it
>>>> and I think that every reference in the YANG should give me all three tiers after IANA in that order perhaps
>>>> IANA; IKEv2 Parameters; Transform Atribute Types; Transform Type 1
>>>
>>> Great. We also considered this as a possible solution after sending our e-mail. Let’s do it.
>>>
>>>> If the syntax needs tweeking, then the RFC Editor will do a good job of that but at present the references are inconsistent in which elements are specified in what order and that is something the RFC Editor probably cannot cope with.
>>>> Authentication Method is a registry so that just needs Group name and Registry name after IANA.
>>>
>>> Ok, good.
>>>>
>>>> Some minor glitches.
>>>> I-D appears twice in the body of the I-D - perhaps document or memo.
>>>
>>> Ok.
>>>
>>>> objetives/objectives/
>>>
>>> Fixed.
>>>> end port number perhaps /must/MUST/
>>>
>>>> and YANG is very good at including such checks with a must ....
>>>> 'If AEAD is used .. where? this occurs in several places and I think that you need to specify the leaf where AEAD will be specified or implied.
>>>
>>> Ok we have changed it to:
>>>
>>> If Authenticated Encryption with Associated Data (AEAD) is used (leaf esp-algorithms/encryption/algorithm-type)
>>> this flag MUST be false."
>>>
>>>> And is it possible to make that a YANG 'must' statement - looking at the IKEv2 registries it is not obvious which are AEAD so that might be more complexity than it is worth.
>>>> 'only available on linux kernels' Um implementation detail, you may get asked to remove that altogether or at least to a Informative Appendix - I would leave it in for now.
>>>
>>> Ok.
>>>
>>>> 'import ietf-i2nsf-ikec' the reference needs to be to a RFC and if it is not yet an RFC then RFC XXXX <title> in both Appendix B and C
>>>
>>> Yes, we have changed that to:
>>>
>>> RFC XXXX: Software-Defined Networking
>>>                 (SDN)-based IPsec Flow Protection.
>>>
>>>> leaf-list pfs-groups could do with a reference - Transform Type 4?
>>>
>>> The leaf list is referring to type pfs-group and we have now
>>>
>>> typedef pfs-group {
>>>              type uint16;
>>>              description
>>>                  "DH groups for IKE and IPsec SA rekey.";
>>>              reference
>>>                  "IANA; Internet Key Exchange V2 (IKEv2) Parameters;
>>> 				Transform Atribute Types; Transform Type 4 -
>>>                  Diffie-Hellman Group Transform IDs.
>>> 				Section 3.3.2 in RFC 7296.";
>>>          }
>>>
>>>
>>>> I am still working my way through the YANG so may have some more comments tomorrow.
>>>
>>> Ok, do not worry we will work in -12 with these comments so far to have a quick response. We can prepare a another version later with the rest of them.
>>>
>>> Thank you very much for your effort.
>>>>
>>>> Tom Petch
>>>>
>>>>> We have realized that we missed to change this, even though we discussed it. We will change it right away in the following way (bold):
>>>>>
>>>>> case rfc822-address-string {
>>>>>       leaf rfc822-address-string {
>>>>>            type string;
>>>>>            description
>>>>>                "Specifies the identity as a
>>>>>                 fully-qualified RFC5322 email
>>>>>                 address string. An example is,
>>>>>                 [e-mail]. The string
>>>>>                 MUST NOT contain any
>>>>>                 terminators e.g., NULL, CR,
>>>>>                 etc.).";
>>>>>             reference
>>>>>                    "RFC 5322.";
>>>>>       }
>>>>> }
>>>>>
>>>>> Btw, we already used in the past “case rfc822-address-string” and “leaf rfc822-address-string” since this is coming from IKEv2 standard. Do you think we should change that name as well?
>>>>>> YANG module references IANA Protocol Numbers which is not in the I-D references
>>>>>
>>>>> We have included the following reference:
>>>>>
>>>>> [IANA-Protocols-Number]
>>>>>                Internet Assigned Numbers Authority (IANA), "Protocol
>>>>>                Numbers", January 2020.
>>>>>
>>>>>
>>>>>>
>>>>>> s.2 boiler plate is out of date
>>>>>
>>>>> What we see is the I-D has the second choice stated in https://www.ietf.org/standards/ids/guidelines/
>>>>>
>>>>> This Internet-Draft is submitted in full conformance with the
>>>>>     provisions of BCP 78 and BCP 79.
>>>>>
>>>>>     Internet-Drafts are working documents of the Internet Engineering
>>>>>     Task Force (IETF).  Note that other groups may also distribute
>>>>>     working documents as Internet-Drafts.  The list of current Internet-
>>>>>     Drafts is at https://datatracker.ietf.org/drafts/current/.
>>>>>
>>>>>     Internet-Drafts are draft documents valid for a maximum of six months
>>>>>     and may be updated, replaced, or obsoleted by other documents at any
>>>>>     time.  It is inappropriate to use Internet-Drafts as reference
>>>>>     material or to cite them other than as "work in progress."
>>>>>
>>>>> Could you refer what is out of date?
>>>>>
>>>>>>
>>>>>> XXXX is standing in for more than one RFC
>>>>>
>>>>> Yes, XXXX has been used because we do not know the future number assigned to our I-D.
>>>>>
>>>>> Also we realized we also included this to refer to crypto-types I-D but this has been solved now in a new version -12 that we are preparing to include your comments. We noticed we can replace the type of rw cert-data?, ca-data*, crl-data? for binary without any problem.
>>>>>
>>>>> |           |     +--rw cert-data?        binary
>>>>> |           +--rw private-key?            binary
>>>>> |           +--rw ca-data*                binary
>>>>> |           +--rw crl-data?               binary
>>>>>
>>>>>>
>>>>>> but the show stopper that makes a proper review of this too costly is the references.  Those to IANA of which there are several I want to pursue.  The I-D reference is to IKEv2 parameters. Sadly, this is a three tier structure and noone agrees on what to call the third tier so I will call it tier3 here.  Top level is Group, as per RFC8126, second level is Registry.  The I-D reference is to the Group only which is fine if the actual reference then specifies the Registry and Tier3 but they never do, usually just Tier3 e.g. Transform Type 3 which makes for a lot of work for the reader, too much for this one.  You have to go hunting in all the second level Registry until you can find a match for the Tier3 identifier. And there are no URL.  If you want an example that I find easy to use, go look at RFC8407 (as usual).
>>>>>
>>>>> You’re right. Could you point the exact part at RFC 8407 with that example? We would really appreciate it.
>>>>>
>>>>> On the other hand, would it be enough to include the URL for Transform Type 3 https://www.iana.org/assignments/ikev2-parameters/ikev2-parameters.xhtml#ikev2-parameters-7 ?
>>>>>
>>>>> (Same for Transform Type 1, Transform Type 4)
>>>>>
>>>>>>
>>>>>> The reference for import of i2nsf-ikec gives a YANG module name; this needs to be the name of the RFC to be
>>>>>
>>>>> Fixed.
>>>>>
>>>>> import ietf-i2nsf-ikec {
>>>>>              prefix nsfikec;
>>>>>              reference
>>>>>                  "RFC XXXX: Software-Defined Networking
>>>>>                 (SDN)-based IPsec Flow Protection.";
>>>>> }
>>>>>
>>>>> We still use XXXX because we do not know the number assigned to the RFC to be.
>>>>>
>>>>>>
>>>>>> The example IPv6 address in the YANG module has :0:0: which is usually just ::
>>>>>
>>>>> Fixed.
>>>>>
>>>>> If you have any further comments, please let us know so we can include them in -12
>>>>>
>>>>> Best Regards.
>>>>>>
>>>>>> And I have some way to go still.
>>>>>>
>>>>>> Tom Petch
>>>>>>
>>>>>> On 22/10/2020 18:39, Rafa Marin-Lopez wrote:
>>>>>>> Dear all:
>>>>>>>
>>>>>>> After receiving a suggestion to make things clearer in the feature ikeless-notification description, we have just uploaded a new version -11 with a minor change to add the following text:
>>>>>>>
>>>>>>> feature ikeless-notification {
>>>>>>>              description
>>>>>>>                  "This feature indicates that the server supports
>>>>>>>                  generating notifications in the ikeless module.
>>>>>>>
>>>>>>>                  To ensure broader applicability of this module,
>>>>>>>                  the notifications are marked as a feature.
>>>>>>>                  For the implementation of ikeless case,
>>>>>>>                  the NSF is expected to implement this
>>>>>>>                  feature.";
>>>>>>>          }
>>>>>>>
>>>>>>> Best Regards.
>>>>>>>
>>>>>>>> Inicio del mensaje reenviado:
>>>>>>>>
>>>>>>>> De: [e-mail]
>>>>>>>> Asunto: New Version Notification for draft-ietf-i2nsf-sdn-ipsec-flow-protection-11.txt
>>>>>>>> Fecha: 22 de octubre de 2020, 15:32:50 CEST
>>>>>>>> Para: "Fernando Pereniguez-Garcia" <[e-mail]>, "Rafael Lopez" <[e-mail]>, "Gabriel Lopez-Millan" <[e-mail]>, "Rafa Marin-Lopez" <[e-mail]>
>>>>>>>>
>>>>>>>>
>>>>>>>> A new version of I-D, draft-ietf-i2nsf-sdn-ipsec-flow-protection-11.txt
>>>>>>>> has been successfully submitted by Rafa Marin-Lopez and posted to the
>>>>>>>> IETF repository.
>>>>>>>>
>>>>>>>> Name:		draft-ietf-i2nsf-sdn-ipsec-flow-protection
>>>>>>>> Revision:	11
>>>>>>>> Title:		Software-Defined Networking (SDN)-based IPsec Flow Protection
>>>>>>>> Document date:	2020-10-22
>>>>>>>> Group:		i2nsf
>>>>>>>> Pages:		92
>>>>>>>> URL:            https://www.ietf.org/archive/id/draft-ietf-i2nsf-sdn-ipsec-flow-protection-11.txt
>>>>>>>> Status:         https://datatracker.ietf.org/doc/draft-ietf-i2nsf-sdn-ipsec-flow-protection/
>>>>>>>> Htmlized:       https://datatracker.ietf.org/doc/html/draft-ietf-i2nsf-sdn-ipsec-flow-protection
>>>>>>>> Htmlized:       https://tools.ietf.org/html/draft-ietf-i2nsf-sdn-ipsec-flow-protection-11
>>>>>>>> Diff:           https://www.ietf.org/rfcdiff?url2=draft-ietf-i2nsf-sdn-ipsec-flow-protection-11
>>>>>>>>
>>>>>>>> Abstract: