Re: [OPSAWG] Benjamin Kaduk's No Objection on draft-ietf-opsawg-nat-yang-16: (with COMMENT)

[Benjamin Kaduk <[e-mail]>]

On Tue, Sep 25, 2018 at 09:27:08PM +0000, [e-mail] wrote:
> Hi Benjamin, 
> 
> Thank you for the comments. 
> 
> Please see inline.

Also inline.

> Cheers,
> Med
> 
> > -----Message d'origine-----
> > De : Benjamin Kaduk [mailto:[e-mail]]
> > Envoyé : mardi 25 septembre 2018 20:31
> > À : The IESG
> > Cc : [e-mail]; Joe Clarke; [e-mail];
> > [e-mail]; [e-mail]
> > Objet : Benjamin Kaduk's No Objection on draft-ietf-opsawg-nat-yang-16: (with
> > COMMENT)
> > 
> > Benjamin Kaduk has entered the following ballot position for
> > draft-ietf-opsawg-nat-yang-16: No Objection
> > 
> > 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-opsawg-nat-yang/
> > 
> > 
> > 
> > ----------------------------------------------------------------------
> > COMMENT:
> > ----------------------------------------------------------------------
> > 
> > Thanks for the easy-to-read document!  I just have a few comments and
> > potential nits
> > I noticed.
> > 
> > It was somewhat interesting to me that basically everything is config rw,
> > including ports and
> > addresses that would normally be assigned internally by the NAT, but I don't
> > see this as
> > problematic.
> 
> [Med] This is a good point. Actually, we are using rw because the same structure is also used for static mappings. That is, the external port and IP address are also provided.  

I had even internalized that, just somehow didn't make the connection.
Thanks for setting me straight.

> > 
> > Section 2.1
> > 
> >                               Considerations about instructing explicit
> >    dynamic means (e.g., [RFC6887], [RFC6736], or [RFC8045]) are out of
> >    scope.  [...]
> > 
> > I'm having trouble parsing this; is it maybe "instructing by explicit
> > dynamic means" or "explicit dynamic mappings"?
> 
> [Med] Changed to "Considerations about instructing by explicit dynamic means". Thanks. 
> 
> > 
> > Section 3
> > 
> > What's the relationship between hold-down-timeout and hold-down-max -- that
> > is, if the maximum number of ports in the pool gets hit, to the oldest
> > ports in the pool get ejected even if they haven't timed out, or what
> > happens?
> > 
> 
> [Med] deallocated ports are added to the hold-down pool till a max is reached; ports are removed from that pool upon the expiry of the hold-down-timeout. New deallocated ports cannot be added if the pool reaches its max. 

Okay, so if the hold-down pool is full and a mapping's expiration timer
expires, does that port immediately become free to use, or does the mapping
persist until there is space in the hold-down pool, or something else?

> > I don't expect this to need to be in the document, but I'm curious what the
> > use case for the all-algs-enable leaf is.
> 
> [Med] This is to allow to enable all "default" ALGs that are widely supported (FTP, RSTP, in particular). This is an optimization as each of the ALGs can be enabled separately. 

Okay.  I guess I don't quite see how this implies the semantics that it
overrides the per-ALG settings, but it's documented well enough that I
don't object to it being this way.

Thanks!

-Benjamin

> > 
> > I may be confused, but is the ordering relationship between low-threshold
> > and high-threshold correct?  From the description it would seem like we
> > need low < high, but I'm reading the text as requiring low >= high.
> > Also, the error-message for that "must" stanza talks about port numbers,
> > not percentage thresholds.
> 
> [Med] Good c	atch. You are completely right. Fixed. 
> 
> > 
> >         container connection-limits {
> >           [...]
> >           list limit-per-protocol {
> >             [...]
> >             leaf limit {
> >               type uint32;
> >               description
> >                 "Rate-limit the number of protocol-specific mappings
> >                  and sessions per instance.";
> > 
> > This is a maximum, not a rate-limit, I think?
> 
>  [Med] Yes. Fixed.
> 
> > 
> > Section A.6
> > 
> >    EAMs may be enabled jointly with statefull NAT64.  This example shows
> >    a NAT64 function that supports static mappings:
> > 
> > nit: "stateful"
> 
> [Med] Fixed. Thanks
> > 
>