Re: [Sidrops] nlnet rp and rsync

[Russ Housley <[e-mail]>]

The repository MUST be available by rsync, and it MAY also be available using other protocols.

A client can use any of the protocols that it wants, but it seems like unnecessary fragility to pick an alternate and then refuse to consider rsync.  This seem counter to a graceful transition, especially if there is a transition from RRDP to RRDP2 in the distant future.

Russ


> On May 11, 2020, at 7:17 AM, Randy Bush <[e-mail]> wrote:
> 
> turning a private discussion public, with consent of course
> 
> it turns out that, to quote tim
> 
>> Routinator has had support for RRDP since version 0.6.0, released in
>> September 2019.
>> 
>> If it finds an RRDP SIA and successfully synchronizes the publication
>> point once, it will whitelist the RRDP SIA URI and it will no longer
>> fall back to rsync where this URI occurs on CA certificates. As long
>> as the synchronization has not been successful it will fall back to
>> rsync.
> 
> to which i then said
> 
>> that seem a potential violation of spec.  if the PP RRDP service then
>> fails, it really MUST fall back to the mandatory to implement rsync.
>> 
>>> As long as the synchronization has not been successful it will fall
>>> back to rsync.
>> 
>> s/As long as/If/
> 
> to which martin said
> 
>> Can you point me to where it says that? Section 3.4.5 of RFC 8182 is
>> pretty indifferent about it.
> 
> to which i said
> 
>> among other places, 6481 s3
>> 
>>     *  The publication repository MUST be available using rsync
>>        [RFC5781] [RSYNC].  Support of additional retrieval mechanisms
>>        is the choice of the repository operator.  The supported
>>        retrieval mechanisms MUST be consistent with the accessMethod
>>        element value(s) specified in the SIA of the associated CA or
>>        EE certificate.
> 
> to which martin responded
> 
>> This talks about what the publication point has to offer, not what a
>> cache has to use. In particular, it says that the publication point
>> musn’t advertise RRDP unless it actually offers RRDP. Which means that
>> as a cache I can rely on RRDP being available.
>> 
>> Forcing a cache to use rsync instead of RRDP is a downgrade attack and
>> should be handled accordingly.
> 
> at which point i said it is time to take this to the wg.  so here we go.
> 
> imiho, 6481 makes it very clear that rsync is currently the mandatory to
> implement protocol (yes i am a coauthor of a draft trying to eventially
> change that).  i contend that this is MTI by both the publisher and by
> the client.
> 
> randy
> 
> _______________________________________________
> Sidrops mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/sidrops