Re: [Int-area] Re draft-daveor-cgn-logging-02/RFC6302

The only problem that I have with this is the use of the word “should” - I hope I’m not splitting hairs here, but I think there is a slight risk of "victim blaming". 

Consider the scenario where the entity with the Internet-facing server (and therefore with the logs) is a victim of some sort of crime. They have the required logs but they weren’t aware that there was a time offset with reference to a global time source. Again, this is something that happens all the time. Interpreted in this context, I think an indication of what they should have been doing might be a bit on the strong side. What do you think? 

What about this weaker-worded alternative:

“If the entity controlling the server is aware that there is an offset required to synchronise with a global time source, it is expected that the offset would be indicated by the entity while the logs were being collected.” 

daveor

> On 9 Apr 2018, at 07:26, <mohamed.boucadair@orange.com> <mohamed.boucadair@orange.com> wrote:
> 
> Hi Dave, 
> 
> What about: 
> 
> "The entity which owns the server should indicate the required offset to synchronize with a global time source."
> 
> Cheers,
> Med
> 
>> -----Message d'origine-----
>> De : Dave O'Reilly [mailto:rfc@daveor.com]
>> Envoyé : samedi 7 avril 2018 16:31
>> À : BOUCADAIR Mohamed IMT/OLN
>> Cc : int-area@ietf.org
>> Objet : Re: [Int-area] Re draft-daveor-cgn-logging-02/RFC6302
>> 
>> Hi Mohamed,
>> 
>> I dont agree with this bit:
>> 
>>> Adjusting the log records to synchronize with a global time source is the
>> responsibility of the entity which owns the server.
>> 
>> I think that both in principle and in practice this
>> synchronisation/correction would be carried out by law enforcement as part of
>> their investigation. There might, I suppose, be an expectation that a server
>> operator would indicate if there was a difference between the times in their
>> logs and a standard time reference but in any case the law enforcement
>> officer is going to have to go through the logs and calibrate the times in
>> the context of whatever matter they are investigating.
>> 
>> The log data plus analysis/calibration would form part of the justification
>> for issuing a subpoena for CGN records (depending on jurisdiction), and the
>> law enforcement officer would have to be able to stand over the grounds for
>> accessing the logs if the request is challenged. If the information being
>> requested is heavily dependent on the accuracy of the times stated in the
>> request, as might be the case if CGN was in use, one could reasonably expect
>> to be asked to justify that the times indicated are accurate (with reference
>> to some sort of time standard) - at which point the law enforcement officer,
>> forensic analyst, or whoever gathered the evidence would need to be able to
>> explain how they concluded that the times in the subpoena were the correct
>> ones. This would presumably include any offset calibration that was carried
>> out, or at least the results of an investigation to confirm that such a
>> calibration was not required.
>> 
>> Also, if a server operator adjusted the times in logs before providing them
>> as evidence, it is not beyond the realm of possibility that the
>> authenticity/integrity of the evidence could be challenged because the log
>> data has been altered since it was recorded.
>> 
>> Regards,
>> daveor
>> 
>>> On 6 Apr 2018, at 08:03, <mohamed.boucadair@orange.com>
>> <mohamed.boucadair@orange.com> wrote:
>>> 
>>> Hi Dave,
>>> 
>>> Glad to see that we are in agreement.
>>> 
>>> I don't think that those sections are needed for the reasons explained in
>> my previous message.
>>> 
>>> One way to avoid misinterpreting your draft as conflicting with existing
>> RFCs is to tweak section 7.4, e.g.:
>>> 
>>> OLD:
>>> 
>>>  There are many reasons why it is may not be possible to record logs
>>>  with reference to a centralised time source (e.g.  NTP).  This could
>>>  include scenarios should as security sensitive networks, or internal
>>>  production networks.  Times MAY OPTIONALLY be recorded with reference
>>>  to a centralised time source (e.g.  NTP) but this is not necessary.
>>>  As long as times are recorded consistently, it should be possible to
>>>  measure the offset from a reference time source if required for the
>>>  purposes of quering records at another source.  This is common
>>>  practice in digital forensics.
>>> 
>>> NEW:
>>> 
>>>  There are many reasons why it may not be possible for servers to record
>> logs
>>>  with reference to a global time source.  This could
>>>  include scenarios such as security sensitive networks, or internal
>>>  production networks. As long as times are recorded consistently, it
>> should be possible to
>>>  measure the offset from a traceable global time source (if required) for
>> the
>>>  purposes of querying records at another source. Adjusting the log records
>> to
>>>  synchronize with a global time source is the responsibility of the entity
>>>  which owns the server.
>>> 
>>> Cheers,
>>> Med
>>> 
>>>> -----Message d'origine-----
>>>> De : Dave O'Reilly [mailto:rfc@daveor.com]
>>>> Envoyé : jeudi 5 avril 2018 16:29
>>>> À : BOUCADAIR Mohamed IMT/OLN
>>>> Cc : int-area@ietf.org
>>>> Objet : Re: [Int-area] Re draft-daveor-cgn-logging-02/RFC6302
>>>> 
>>>> Hi Mohamed,
>>>> 
>>>> Thanks for your mail.
>>>> 
>>>> I agree with you.
>>>> 
>>>> The only reason I put these sections in here was because the IESG conflict
>>>> review indicated a conflict between this document and the other two RFCs
>>>> mentioned (Ref: https://datatracker.ietf.org/doc/conflict-review-daveor-
>> cgn-
>>>> logging/). In an effort to reconcile the feedback received with the
>> content
>>>> of draft-daveor-cgn-logging, I added these sections.
>>>> 
>>>> Perfectly happy to remove them if that is the way the consensus emerges.
>>>> 
>>>> daveor
>>>> 
>>>> 
>>>>> On 5 Apr 2018, at 15:24, <mohamed.boucadair@orange.com>
>>>> <mohamed.boucadair@orange.com> wrote:
>>>>> 
>>>>> Hi Dave,
>>>>> 
>>>>> I have a comment about the proposed update to RFC 6269 (the same comment
>>>> applies for RFC6302, though).
>>>>> 
>>>>> Actually, the proposed NEW text will require an extra effort to align
>>>> timestamps among the server which maintains the logs, the authorities that
>>>> relay an abuse claim, and the provider who manages the CGN. That extra
>> effort
>>>> has to be done by the entity managing the log server.
>>>>> 
>>>>> From that standpoint, the proposed NEW text is no more than another
>> example
>>>> of "Accurate time-keeping"...which IMHO does not justify an update to the
>>>> 6269.
>>>>> 
>>>>> Cheers,
>>>>> Med
>>>>> 
>>>>>> -----Message d'origine-----
>>>>>> De : Int-area [mailto:int-area-bounces@ietf.org] De la part de Dave
>>>> O'Reilly
>>>>>> Envoyé : mercredi 4 avril 2018 22:26
>>>>>> À : int-area@ietf.org
>>>>>> Objet : Re: [Int-area] Re draft-daveor-cgn-logging-02/RFC6302
>>>>>> 
>>>>>> Dear all,
>>>>>> 
>>>>>> Further to my email below, I have revised draft-daveor-cgn-logging and
>>>>>> revision -03 is now available. I have restructured the content into the
>>>> form
>>>>>> of recommendations.
>>>>>> 
>>>>>> Here’s the link: https://tools.ietf.org/html/draft-daveor-cgn-logging-03
>>>>>> 
>>>>>> I have also included, at sections 7.6 and 7.7, proposed amendments to
>>>> RFC6302
>>>>>> and RFC6269 respectively.
>>>>>> 
>>>>>> Regards,
>>>>>> daveor
>>>>>> 
>>>>>>> On 20 Mar 2018, at 13:45, Dave O'Reilly <rfc@daveor.com> wrote:
>>>>>>> 
>>>>>>> Dear all,
>>>>>>> 
>>>>>>> further to presenting at IETF-101 yesterday I wanted to send a follow
>> up
>>>>>> email to see if there is interest in working on a new best current
>>>> practice
>>>>>> for logging at internet-facing servers.
>>>>>>> 
>>>>>>> I hope I adequately presented the reasons why I think there needs to be
>>>>>> some revision of the recommendations of RFC6302 and that there is some
>>>>>> additional points to be considered in draft-daveor-cgn-logging-02.
>>>>>>> 
>>>>>>> The current version of the document (draft-daveor-cgn-logging-02)
>>>> contains
>>>>>> recommendations, but it is not really in the form of a BCP. If there is
>>>>>> interest, I would like to suggest, in the first instance at least, that
>> I
>>>>>> prepare a new version of the document, structured in the form of a BCP
>>>> with a
>>>>>> set of recommendations for discussion.
>>>>>>> 
>>>>>>> Any feedback would be appreciated.
>>>>>>> 
>>>>>>> Thanks and best regards,
>>>>>>> daveor
>>>>>>> 
>>>>>>> _______________________________________________
>>>>>>> Int-area mailing list
>>>>>>> Int-area@ietf.org
>>>>>>> https://www.ietf.org/mailman/listinfo/int-area
>>>>>> 
>>>>>> _______________________________________________
>>>>>> Int-area mailing list
>>>>>> Int-area@ietf.org
>>>>>> https://www.ietf.org/mailman/listinfo/int-area
>>> 
> 