Re: [sidr] Alexey Melnikov's Discuss on draft-ietf-sidr-delta-protocol-07: (with DISCUSS and COMMENT)

[Alexey Melnikov <[e-mail]>]

Hi Tim,
(Top post)

I think your latest suggested text is good enough, so polishing it more is not going to improve the document in a significant way. So please post a new version with it.

Thank you,
Alexey

> On 8 Mar 2017, at 09:29, Tim Bruijnzeels <[e-mail]> wrote:
> 
> Hi Alexey
> 
>> On 07 Mar 2017, at 16:15, Alexey Melnikov <[e-mail]> wrote:
>> 
>> Hi Tim,
>> 
>>> On Tue, Mar 7, 2017, at 01:38 PM, Tim Bruijnzeels wrote:
>>> Hi all,
>>> 
>>>> On 02 Mar 2017, at 12:04, Alexey Melnikov <[e-mail]> wrote:
>>>> 
>>>> Alexey Melnikov has entered the following ballot position for
>>>> draft-ietf-sidr-delta-protocol-07: Discuss
>>>> 
>>>> When responding, please keep the subject line intact and reply to all
>>>> email addresses included in the To and CC lines. (Feel free to cut this
>>>> introductory paragraph, however.)
>>>> 
>>>> 
>>>> Please refer to https://www.ietf.org/iesg/statement/discuss-criteria.html
>>>> for more information about IESG DISCUSS and COMMENT positions.
>>>> 
>>>> 
>>>> The document, along with other ballot positions, can be found here:
>>>> https://datatracker.ietf.org/doc/draft-ietf-sidr-delta-protocol/
>>>> 
>>>> 
>>>> 
>>>> ----------------------------------------------------------------------
>>>> DISCUSS:
>>>> ----------------------------------------------------------------------
>>>> 
>>>> I would be happy to ballot Yes on this document, as it is well written
>>>> and is a useful piece of work. However I have one issue (and a few minor
>>>> comments) that I would like to DISCUSS before doing so:
>>>> 
>>>> In Section 5.3 the document says:
>>>> 
>>>>  It is RECOMMENDED that Relying Parties and Publication Servers
>>>> follow
>>>>  the Best Current Practices outlined in [RFC7525] on the use of HTTP
>>>>  over TLS (HTTPS) [RFC2818].
>>>> 
>>>> RFC 7525 is referencing RFC 6125 for server hostname validation.
>>>> Unfortunately this is not detailed enough to perform hostname validation,
>>>> because reference to RFC 6125 requires specifying answers to every
>>>> question in section 3 of RFC 6125. (And there is no generic RFC that
>>>> specifies how this is done for protocols using HTTP.) One example of how
>>>> this might look like is in Section 9.2 of
>>>> <https://datatracker.ietf.org/doc/draft-ietf-sidr-rpki-rtr-rfc6810-bis/?include_text=1>.
>>>> For your convenience the relevant text is pasted below:
>>>> 
>>>>  Routers MUST also verify the cache's TLS server certificate, using
>>>>  subjectAltName dNSName identities as described in [RFC6125], to
>>>> avoid
>>>>  man-in-the-middle attacks.  The rules and guidelines defined in
>>>>  [RFC6125] apply here, with the following considerations:
>>>> 
>>>>     Support for DNS-ID identifier type (that is, the dNSName identity
>>>>     in the subjectAltName extension) is REQUIRED in rpki-rtr server
>>>>     and client implementations which use TLS.  Certification
>>>>     authorities which issue rpki-rtr server certificates MUST support
>>>>     the DNS-ID identifier type, and the DNS-ID identifier type MUST
>>>> be
>>>>     present in rpki-rtr server certificates.
>>>> 
>>>>     DNS names in rpki-rtr server certificates SHOULD NOT contain the
>>>>     wildcard character "*".
>>>> 
>>>>     rpki-rtr implementations which use TLS MUST NOT use CN-ID
>>>>     identifiers; a CN field may be present in the server
>>>> certificate's
>>>>     subject name, but MUST NOT be used for authentication within the
>>>>     rules described in [RFC6125].
>>>> 
>>>> The only thing missing from the above is explicit mentioning that SRV-ID
>>>> and URI-ID are not used. (I think the same should apply to your
>>>> document.)
>>> 
>>> Considering that we also say:
>>> 
>>>   Relying Party tools SHOULD log any TLS certificate or host name
>>>   validation issues thus found, so that an operator can investigate the
>>>   cause.  However, such validation issues are often due to
>>>   configuration errors, or a lack of a common TLS trust anchor.  In
>>>   these cases it is better if the Relying Party retrieves the signed
>>>   RPKI data regardless, and performs validation on it.  Therefore
>>>   Relying Party MUST continue to retrieve the data in case of errors.
>>>   The Relying Party MAY choose to log encountered issues only when
>>>   fetching the notification update file, but not when it subsequently
>>>   fetches snapshot or delta files from the same host.  Furthermore the
>>>   Relying Party MAY provide a way for operators to accept untrusted
>>>   connections for a given host, after the cause has been identified.
>>> 
>>> 
>>> And because in practice Relying Party software will use standard software
>>> libraries to do retrieval and verification, and it may be hard or even
>>> impossible to configure these libraries to do the verification as
>>> described.. would you agree with the following? Essentially taking your
>>> suggested lead, but never exceeding "SHOULD" in the "how" of the TLS
>>> certificate and host name validation that itself is a SHOULD, i.e.:
>> 
>> I think it would be clearer to say that *if* validation is done, it will
>> use the following MUST/MUST NOTs. (And if validation fails, you log as
>> you recommend elsewhere.) Otherwise it would not be very clear when it
>> is Ok to violate various SHOULD/SHOULD NOTs.
> 
> I see your point regarding clarity, but my main point is that I believe it would be overkill to have MUSTs. It may not be possible to enforce this behaviour in standard libraries of the language of choice for Relying Party implementations (currently Java (RIPE NCC), Python (rcynic) and C I believe for RIPSTR). And since we have RPKI object security I don't believe it would be a big issue if for example a wildcard certificate is used.
> 
> I really do not want to have to implement our own https client library for this. It would require a lot of work and maintenance, and on the whole it would make our implementation brittle. I want to be able to use one of the widely used and tested standard libraries for this, even if they violate a SHOULD. Of course I am willing to see whether we can tweak behaviour of said libraries, but again this may not possible.
> 
> If you agree, would it help if we moved the text around a bit (so that the security motivation comes first) and add this explanation explicitly at the end:
> 
> 4.3.  HTTPS considerations
> 
>    Note that a Man-in-the-Middle (MITM) cannot produce validly signed
>    RPKI data, but can perform withhold or replay attacks targeting an
>    Relying Party, and keep the Relying Party from learning about changes
>    in the RPKI.  Because of this Relying Parties SHOULD do TLS
>    certificate and host name validation when they fetch from an RRDP
>    Repository Server.
> 
>    Relying Party tools SHOULD log any TLS certificate or host name
>    validation issues found, so that an operator can investigate the
>    cause.  However, such validation issues are often due to
>    configuration errors, or a lack of a common TLS trust anchor.  In
>    these cases it is better if the Relying Party retrieves the signed
>    RPKI data regardless, and performs validation on it.  Therefore
>    Relying Party MUST continue to retrieve the data in case of errors.
>    The Relying Party MAY choose to log encountered issues only when
>    fetching the notification update file, but not when it subsequently
>    fetches snapshot or delta files from the same host.  Furthermore the
>    Relying Party MAY provide a way for operators to accept untrusted
>    connections for a given host, after the cause has been identified.
> 
>    It is RECOMMENDED that Relying Parties and Repository Servers follow
>    the Best Current Practices outlined in [RFC7525] on the use of HTTP
>    over TLS (HTTPS) [RFC7230].  Relying Parties SHOULD do TLS
>    certificate and host name validation using subjectAltName dNSName
>    identities as described in [RFC6125].  The rules and guidelines
>    defined in [RFC6125] apply here, with the following considerations:
> 
>    o  Relying Parties and Repository Servers SHOULD support the DNS-ID
>       identifier type.  The DNS-ID identifier type SHOULD be present in
>       Repository Server certificates.
> 
>    o  DNS names in Repository Server certificates SHOULD NOT contain the
>       wildcard character "*".
> 
>    o  A CN field may be present in Repository Server certificates's
>       subject name, but SHOULD NOT be used for authentication within the
>       rules described in [RFC6125].
> 
>    o  This protocol does not require the use of SRV-IDs.
> 
>    o  This protocol does not require the use of URI-IDs.
> 
>    Note however that this validation is done on a best effort basis, and
>    serves to highlight potential issues, but RPKI object security does
>    not depend on this.  Therefore Relying Parties MAY deviate from the
>    validation steps listed above, in particular in case standard widely
>    used software libraries in the language of the Relying Party software
>    implementation do not support and/or cannot be configured to function
>    this way.
> 
> 
> 
> 
> 
> 
> 
> 
> Thanks
> Tim
> 
> 
> 
>> 
>>>   Note that a Man-in-the-Middle (MITM) cannot produce validly signed
>>>   RPKI data, but can perform withhold or replay attacks targeting an
>>>   Relying Party, and keep the Relying Party from learning about changes
>>>   in the RPKI.  Because of this Relying Parties SHOULD do TLS
>>>   certificate and host name validation when they fetch from an RRDP
>>>   Repository Server, using subjectAltName dNSName identities as
>>>   described in [RFC6125].  The rules and guidelines defined in
>>>   [RFC6125] apply here, with the following considerations:
>>> 
>>>   o  Relying Parties and Repository Servers SHOULD support the DNS-ID
>>>      identifier type. The DNS-ID identifier type SHOULD be present in
>>>      Repository Server certificates.
>>> 
>>>   o  DNS names in Repository Server certificates SHOULD NOT contain the
>>>      wildcard character "*".
>>> 
>>>   o  A CN field may be present in Repository Server certificates's
>>>      subject name, but SHOULD NOT be used for authentication within the
>>>      rules described in [RFC6125].
>>> 
>>>   o  This protocol does not require the use of SRV-IDs.
>>> 
>>>   o  This protocol does not require the use of URI-IDs.
>>> 
>>>> 
>>>> 
>>>> ----------------------------------------------------------------------
>>>> COMMENT:
>>>> ----------------------------------------------------------------------
>>>> 
>>>> In 3.2: HTTPS reference is out-of-date.
>>> 
>>> updated to RFC7230
>>> 
>>>> SHA-256 needs a reference.
>>> 
>>> ok, added a normative reference (same as in the publication protocol
>>> document):
>>> 
>>>   [SHS]      National Institute of Standards and Technology, "Secure
>>>              Hash Standard", March 2012,
>>>              <http://csrc.nist.gov/publications/fips/fips180-4/
>>>              fips-180-4.pdf>.
>