Hi Martin,

I'm fine with your answers and proposed changes.

> Hmm, AFAICT, no special config is needed for mDNS to work?  So if
> the device supports mDNS, it will be used for .local. names, right?

I think that's the main point, but RFC6762 does say this:
  "The ".local."
   suffix could appear in the search list because the user manually
   configured it, or because it was received via DHCP [RFC2132] or via
   any other mechanism for configuring the DNS search list."
That means it would be quite legitimate to configure it with NETMOD.
So probably you can ignore my comment ;-).

(There's also a more general question: what does the host do if it
receives conflicting config via DHCP and NETMOD? I guess you have a
general answer to that.)

> Or are you suggesting that this model should support configuration of
> mDNS as well?

As far as I know, mDNS is intended to be a zero-configuration
mechanism, so I think the answer is "No". It might be worth checking
with somebody in the DNSSD WG though.

Regards
   Brian

On 13/01/2014 00:16, Martin Bjorklund wrote:
> Hi,
> 
> Thank you for this review.  Comments inline.
> 
> Brian E Carpenter <[email protected]> wrote:
>> I am the assigned Gen-ART reviewer for this draft. For background on
>> Gen-ART, please see the FAQ at
>> <http://wiki.tools.ietf.org/area/gen/trac/wiki/GenArtfaq>.
>>
>> Please resolve these comments along with any other Last Call comments
>> you may receive.
>>
>> Document: draft-ietf-netmod-system-mgmt-10.txt
>> Reviewer: Brian Carpenter
>> Review Date: 2014-01-12
>> IETF LC End Date: 2014-01-22
>> IESG Telechat date:
>>
>> Summary:  Almost ready
>> --------
>>
>> Comments:
>> ---------
>>
>> I haven't checked the yang syntax in detail. At first glance,
>> it looks good.
>>
>> Minor Issues:
>> ------------
>>
>> "2.1.  System Identification
>>
>>    There are many common properties used to identify devices, operating
>>    systems, software versions, etc. that need to be supported in the
>>    system data module.  These objects are defined as operational state
>>    data and the information returned by the server is intended to be
>>    specific to the device vendor."
>>
>> Does "server" mean the NETCONF server?
> 
> Yes.  We usually include the text below in the Terminology section,
> and we should do that in this document as well.  We missed it.
> 
>    The following terms are defined in [RFC6241] and are not redefined
>    here:
> 
>    o  client
> 
>    o  configuration data
> 
>    o  server
> 
>    o  state data
> 
> I will add this text.
> 
>> "2.2.  System Time Management
>>
>>    The management of the date and time used by the system need to be
>>    supported.  Use of one or more NTP servers to automatically set the
>>    system date and time need to be possible.  Utilization of the
>>    Timezone database [RFC6557] also need to be supported.  It should be
>>    possible for the server, as well as clients, to configure the system
>>    to use NTP."
>>
>> Again, does this mean the NETCONF server? Also, which clients? This is
>> the first time the document mentions clients.
> 
> See above.
> 
>> "2.3.  User Authentication
>>
>> ...It should be possible for the server, as well as
>>    clients, to configure the system authentication properties."
>>
>> Same comment as for 2.2. Also, how does the network administrator enforce
>> a specific strength of authentication? It seems as though the clients can
>> choose to downgrade security.
> 
> I agree that this sounds a bit weird - servers do not configure
> anything spontaneously.  Clients can not in general choose to
> downgrade security.  It simply means that the data model should make
> it possible to configure user authentication.  Access control must be
> used to control which clients are authorized to do so.
> 
> NEW:
> 
>    It should be possible for clients to configure the system
>    authentication properties.
> 
> 
>> "2.4.  DNS Resolver
>>
>>    The configuration of the DNS resolver within the system containing
>>    the NETCONF server is required to control how domain names are
>>    resolved."
>>
>> This is very hard to parse and could mean several things. Does it
>> mean that it MUST be possible to configure the DNS resolver?
> 
> Note that this is the "Objectives" section.  It means that an
> objective is that the data model should allow for dns resolver
> configuration.
> 
> Would it be more clear to say 
> 
>     The configuration of the DNS resolver within the system containing
>     the NETCONF server is required in order to control how domain
>     names are resolved.
> 
> 
>> (And I wonder what happens if the device is running mDNS?)
> 
> Hmm, AFAICT, no special config is needed for mDNS to work?  So if
> the device supports mDNS, it will be used for .local. names, right?
> 
> Or are you suggesting that this model should support configuration of
> mDNS as well?
> 
>> "3.6.  System Control
>>
>>    The following operations are defined:
>>
>>       set-current-datetime
>>       system-restart
>>       system-shutdown"
>>
>> My reaction is that this is a very limited set of operations.
> 
> Yes.  We tried to come up with a useful limited set of operations (and
> config / oper data) that can be implemented on almost all devices.
> 
>> For example
>> how about system standby (enter power-saving mode)? Yes, I know that
>> creates the problem of how to wake the system up again, but there are
>> solutions to that problem (e.g. polling). I can imagine many vendor-specific
>> extensions in this area, too.
> 
> There are probably lots of other operations that can be useful.
> However, we need to start somewhere, and more operations can be added
> over time.  In general, we expect vendor-specific extensions to both
> the supported set of operations, and the config and oper data.
> 
> 
> /martin
> 
_______________________________________________
Gen-art mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/gen-art

Reply via email to