Re: [nfsv4] Alissa Cooper's Discuss on draft-ietf-nfsv4-rfc3530-migration-update-07: (with DISCUSS and COMMENT)

[David Noveck <[e-mail]>]

> (And my bad for not noticing this before
> the publication of RFC 7530, since they would have applied there as
> well.)

You're not alone.  These same issues apply to RFCs 3530 and 5661
as well.

> As best I can tell, embedding the client's IP address in the client id is
> not a requirement (but I admit to not fully understanding NFS!). So why
> is it suggested that the client's network address be embedded in the
> client id?

I think the reason is primarily historical inertia.The idea that
the IP address is a reasonable basis for this goes back to RFC3010.

> It seems to me that to avoid all of the issues above, perhaps a better
> example to provide would be hash(some client secret, some source of
> uniqueness, server identity).  That gives the client id a good chance of
> being unique without exposing other identifiers related to the client.
> And then if the existing guidance about saving and re-using the value is
> followed, it won't be necessary to try to shoehorn in an old IP address
> or persistent device identifiers when the clientid needs to be
> regenerated.

I've already modified my text to include the idea of a hash based on
a suggestion in Elwyn Davies' gen-art review.

I'll make some further changes to bring the client id string guidance into
closer alignment with the actual requirements for the client id string
before
I submit -08.

On Tue, Jan 19, 2016 at 5:38 PM, Alissa Cooper <[e-mail]> wrote:

> Alissa Cooper has entered the following ballot position for
> draft-ietf-nfsv4-rfc3530-migration-update-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-nfsv4-rfc3530-migration-update/
>
>
>
> ----------------------------------------------------------------------
> DISCUSS:
> ----------------------------------------------------------------------
>
> Per my comment below, it seems like there are some fairly significant
> privacy considerations related to the choice of how the client id is
> constructed. I think these need to be described in the document so
> implementors are aware of them. (And my bad for not noticing this before
> the publication of RFC 7530, since they would have applied there as
> well.)
>
>
> ----------------------------------------------------------------------
> COMMENT:
> ----------------------------------------------------------------------
>
> My comments are about the example given in Section 4.9.
>
> I understand from 4.2 that the requirements for the construction of the
> client id are:
>
> 1) It should be unique per client.
> 2) It should persist through reboots.
>
> As best I can tell, embedding the client's IP address in the client id is
> not a requirement (but I admit to not fully understanding NFS!). So why
> is it suggested that the client's network address be embedded in the
> client id?
>
> There are also a few privacy-related issues with the example:
>
> - Some clients will change their IP addresses over time to avoid being
> tracked. By suggesting that some prior address be used in the client id,
> there is an implied requirement on the client to maintain a history of
> previously used addresses, which could be exploited for tracking
> purposes.
>
> - Permanent device identifiers (such as a serial numbers) should not be
> embedded in a client id on the wire, again to avoid facilitating tracking
> by any other party that knows the serial number.
>
> It seems to me that to avoid all of the issues above, perhaps a better
> example to provide would be hash(some client secret, some source of
> uniqueness, server identity). That gives the client id a good chance of
> being unique without exposing other identifiers related to the client.
> And then if the existing guidance about saving and re-using the value is
> followed, it won't be necessary to try to shoehorn in an old IP address
> or persistent device identifiers when the clientid needs to be
> regenerated.
>
>
>