[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: WG Last Call: a batch of draft-ietf-v6ops-ipv4survey-*01 documents
On Wed, 3 Sep 2003, Pekka Savola wrote:
> Thanks for a *very* extensive review of the ops document. I'll
> try to respond to your points somehow inline..
>
> I've included all of your comments which may need addressing in
> the quote, so that it's easier for the document editor to go
> through the spec.
I'm going to trim this down just to the stuff where we still
disagree.
> Three generic comments:
> 1) (now) HISTORIC documents being included.
>
> Unfortunately, the survey was executed in a non-optimal
> method, and these historic statements still appear here, due to
> examining the RFC's in the "vacuum". (The best way to do the
> analysis in the first place would have been to skip those which
> have been obsoleted, made historic, etc., but we lack the time
> machine.. :-)
Actually, that's not entirely true ... some of the reclassifications
(in particular for RFC 1157, SNMPv1, and RFC 1901, SNMPv2c) postdate
the original work.
> I'd like to go through this with a "minimum update" policy, and
> we state that the document is historic in section 7.
A "minimum update" policy that precludes putting in the most recent
versions of important specifications is not a good policy, IMHO,
as it would seriously compromise the utility and credibility of the
document. If I see a lot of stuff in it that I KNOW is obsolete,
then why should I (as a non-v6ops type) pay attention to it?
> Unless there is strong feeling that we should really throw away
> these historic documents completely (implies that we should look
> at which documents made them historic, replacement documents
> etc.), I'd keep them here for now.
I find it hard to accept your proposal in its entirety, but I will
agree to this much: I won't ask the editor to re-survey the status
of all the RFCs. I will, however, continue to push to have
something done about the specific document that Juergen Schoenwalder
and I have explicitly identified as HISTORIC documents.
Now, my original reason for asking that the documents identified as
HISTORIC just be dropped was to make minimum work for the readers of
the document. And since the document needs to be updated anyway, it
seems to be a fairly straightforward matter for the editor to just
delete those sections.
On the other hand, if the WG feels that there is some value in
keeping the parts of sections 3, 4, and 5 for the HISTORIC specs
that are still widely deployed then I won't disagree. But it will
be necessary to make sure the text is accurate (I will try to help
in further comments below) and then in Section 7 you must replace
ALL the verbiage for such specs with a statement like this:
This document has been reclassified as HISTORIC. It has
been replaced by <insert document list here>.
If there are cases where there is no replacement then the second
sentence can be left out.
But as for things that are not now and have never been widely
deployed (such as te DNS MIBs), I don't see any reason to keep them
in the best strategy; and even if you do you should not bother
wasting the reader's time by spelling out their deficiencies. You
should leave such things decently buried in the document graveyard
and not exhume the corpses for forensic examination.
> 1.b) new RFC's since about RFC3200 (in most part, obsoleting or
> making historic those that are still analyzed) have not been
> taken into consideration.
>
> The analysis was Phil's work, who has retired. We don't have
> resources anymore to do more substantial work anymore, so we
> prefer to keep doing minimum updates.
Allow me to propose this compromise: fix the sections on HISTORIC
documents, or documents that have been updated by newer RFCs, where
the reviewers have identified a specific fix, but accept no
obligation to re-survey all the RFCs. Can you and the editor live
with that? I think it's understood that the subject matter is too
much of a moving target for the survey to remain cmpletely accurate
for very long.
> 2) protocol/specification. I totally agree that we should refer the
> RFC's by "specification" because that's generic enough to cover any RFC.
>
> 3) MIB acronyms like TRANS-MIB in:
>
> 3.9 RFC 3417 Transport Mappings for Version 2 of the Simple Network
> Management Protocol (SNMPv2) (TRANS-MIB)
>
> .. these do not appear in the original RFC's, but are invented by
> RFC-editor for some RFC indexes. My opinion is that we should get
> rid of these bogus abbreviations completely, but I would also be
> OK with it if they're changed to be correct. Whatever works for
> you..
Getting rid of the acronyms and relying on RFC title and number
alone would suit me just fine. But if they remain the acronyms must
be the right ones (MIB module names in the case of MIB modules).
Note that in some cases an acronym appears twice, once in the
original title, and the once again and the MIB or protocol name;
in those cases remove only the second occurrence.
> On Sun, 31 Aug 2003, C. M. Heard wrote:
> [...]
> > There are some important technical specifications that do not fall into
> > any of these classifications. For instance, RFC 1215, which is part of
> > the (obsolescent) SMIv1 specification, is an Informational RFC. And RFC
> > 3584 (the coexistence document that replaces RFC 2576) is a BCP. I'm
> > not aware of any other such specs, but some may exist. Certainly it
> > behooves us to include at least those two in the survey. One possible
> > way to handle RFC 1215 would be to include it in Sections 3 and 7 along
> > with its companion SMIv1 documents (RFCs 1155 and 1212). I'm not really
> > sure what to do for RFC 3584; maybe a separate section for BCPs would be
> > appropriate. One thing for sure, it would not be a good idea to lump
> > all non-standards-track stuff (BCP + informational + experimental)
> > together.
>
> The problem here is that it would be difficult for *us* (read: those
> non-experts in the subject matter) to decide what is relevant and what is
> not. The analysis was already done by Phil, and we don't have resources
> to extend it.
OK, I understand that.
> But I think we can address this situation in practical terms. Where
> applicable, add notions about these documents in the "analysis", section
> 7, like:
>
> 7.3.18 Coexistence of SNMP v1, v2, & v3 (RFC 2576)
>
> There are no real issues that can be resolved.
>
> ==> add a sentence like:
>
> The document is obsoleted by RFC 3584.
>
> (or something else, if RFC3584 addresses the concerns)
>
> .. that way, we could have somekind of a "forward pointer" even to
> informational/BCP documents which may be relevant.
>
> Would this satisfy your concern? If so, could you submit some wording
> that would be appropriate for these cases?
I'm not too keen on the specific suggestion above. But, given your
lack of resources, I can accept that you would only address the
specific documents mentioned above (namely RFCs 1212, 1215, and
3584). I can also accept the following editorial strategy:
mention RFC 1215 in conjunction with 1155 and 1212 (since together
they form a package, and really are part of the same de-facto
standard), replace all mentions of RFC 2576 with RFC 3584, move RFC
3584 to Section 3, and re-title Section 3 as follows:
3.0 Full Standards and BCPs
That would entail an addition to the lead-in text of that section:
> > | Full Internet Standards (most commonly simply referred to as
> > | "Standards") are fully mature protocol specification that are widely
> > | implemented and used throughout the Internet.
The OPS area has one technical specification (RFC 3584) which is
classified as a Best Current Practice, or BCP. That is another
"terminal" classification. It was used for this specfication
because it was deemed mature and unlikely to see further evolution,
but did not meet the formal requirements for Full Standard.
Suggested text for the other stuff will appear below.
> > | 3.1 RFC 1157 Simple Network Management Protocol
> > |
> > | Beginning in Section 3.2.6.3.2 atTable Object Type Names thru the rest
> > | of Section 3 there are numerous references to the use of IPv4 addresses
> > | as part of OIDs.
OK, if we want to keep this one (which would be justified because of its
wide deployment) then make this fix:
s/as part of OIDs/for identification of object instances/
Technically, the use of the term "OIDs" is correct here, but I much
prefer the more precise language.
> > | Section 4. Protocol Specification specifies the format of an SNMP
> > | packet which uses the overall format of:
> > |
> > | RFC1157-SNMP DEFINITIONS ::= BEGIN
> > | IMPORTS
> > | ObjectName, ObjectSyntax, NetworkAddress, IpAddress, TimeTicks
> > | FROM RFC1155-SMI;
> > |
> > |
> > | Section 4.1.3.1. Example of Table Traversal has many uses of IPv4
> > | addresses in its example of table transversal.
> > |
> > | Section 5. Definitions reiterates the use of IPv4 addresses.
> > |
> > | RFC1157-SNMP DEFINITIONS ::= BEGIN
> > |
> > | IMPORTS
> > | ObjectName, ObjectSyntax, NetworkAddress, IpAddress, TimeTicks
> > | FROM RFC1155-SMI;
The above is OK.
> I'd like to go through this with a "minimum update" policy, and we state
> that the document is historic in section 7.
>
> Is that good enough?
For this one, yes.
> > | 3.2 RFC 1155 Structure of Management Information
> > |
> > | Section 3.2.3.2. IpAddress defines the following:
> > |
> > | This application-wide type represents a 32-bit internet address. It
> > | is represented as an OCTET STRING of length 4, in network byte-order.
> > |
> > | There are several instances of the use of this definition in the rest of
> > | the document.
> >
> > There should also be a subsection for RFC 1212, and perhaps also RFC
> > 1215, since RFCs 1155 and 1212 form STD 16, and together with RFC
> > 1215 define SMIv1.
>
> We could add here a sentence like:
>
> Note that RFC 1212 and RFC 1215, which have been omitted from analysis,
> together with this specification define SMIv1. The same issues apply
> to all of them.
>
> would that be OK? Another piece of text?
I can live with that, if we fix the analysis sections for 1157/1155
properly. See below.
> (I'm purposely trying to "wave away" the need for formal analysis
> sections..)
I won't quite let you get away with that but I will write the analysis
for you :)
> > | 3.4 RFC 1643 Definitions of Managed Objects for the Ethernet-like
> > | Interface Types
> > |
> > | There are no IPv4 dependencies in this specification.
> >
> > In [a] recent protocol action [...] the IESG reclassified
> > RFC 1643 to HISTORIC [...]. I therefore suggest that RFC
> > 1643 be dropped from this survey.
>
> Due to the way survey is done, I'd like to keep it as is (minimum
> changes)
I don't understand why that's a consideration
> [but] we could add a sentence like:
>
> Moreover, RFC 1643 has been superseded by other MIBs and reclassified
> to HISTORIC.
>
> If that's what you'd prefer?
I do not like this proposal. I know that you have not had the time
to read draft-ietf-hubmib-1643-to-historic-01.txt, but the gist of
it is that the technical content of RFC 1643 has been subsumed by
RFC 2665 (actually by its predecessors), which is already part of
the survey. I don't understand why it is so hard to just drop RFC
1643 from the survey.
However, if you insist on keeping it, could you rewrite the while
section like this:
3.4 RFC 1643 Definitions of Managed Objects for the Ethernet-like
Interface Types
This specification has been reclassified as an HISTORIC
document and is not considered in this discussion.
> > There is no mention of RFC 2580 (which is also an OPS Area Internet
> > Standard). I don't think there are any IPv4 dependencies in RFC 2580,
> > but it should be mentioned here for sake of completeness.
>
> It's PS, so I guess it should be included, with like:
You mean FS
> 3.6 RFC 2580 Conformance Statements for SMIv2
>
> There are no IPv4 dependencies in this specification.
>
> .. if you agree that that is a true statement?
Sounds OK to me. Note that the section number will need to change.
> > RFCs 3411, 3412, 3413, 3414, and 3415 are also Internet Standards;
> > they replace RFCs 2571, 2572, 2573, 2574, and 2575. Please move
> > the sections for those last five RFCs here and update the RFC numbers.
>
> I don't think the editor should have moved RFC3416 etc. here in the first
> place, but as it has already been done, I see no harm doing the same for
> those five other RFCs either.
I do think that the editor should have moved RFC 3416 stuff here :-)
> > | 3.9 RFC 3417 Transport Mappings for Version 2 of the Simple Network
> > | Management Protocol (SNMPv2) (TRANS-MIB)
...
> > s/TRANS-MIB/SNMPv2-TM/
>
> I do not know why those (TRANS-MIB)'s are there in the first place, but in
> case there is a strong reason to keep them there, I vote for removing them
> completely. What we want is that the title here is identical with the RFC
> title.
>
> My suggestion: remove all of (*-MIB) from the document. Would that be
> good for you?
As I said up above, using the exact RFC title with no extra adornment
works for me.
...
> > Wordsmith the last paragraph:
> >
> > Section 8.1, "Usage Example," also contains examples which uses IPv4
> > addresse, but it has no significance in the operation of the
> > specification.
>
> There is typo above, but in practice OK.
>
> s/IPv4 addresse/an IPv4 address/ ?
Yep. Sorry :-(
... skipping down to the proposed standards ...
> > | 5.010 RFC 1441 Introduction to version 2 of the Internet-standard
> > | Network Management Framework (SNMPv2)
> > |
> > | There are no IPv4 dependencies in this protocol.
> >
> > This document is HISTORIC and should be removed from the survey
> > (otherwise, I'd say s/protocol/specification/)
>
> I'd rather keep it in, if you prefer, just stating something like:
>
> Moreover, the document has been reclassied HISTORIC.
No. Absolutely not. This document has been HISTORIC since 1996 (when
1901-1907 appeared) and should never have been part of the original
survey. I repeat again: why do you insist on keeping this stuff?
It does not help your readers or your editor.
> > | 5.018 RFC 1515 Definitions of Managed Objects for IEEE 802.3
> > | Medium Attachment Units (MAUs)
> > |
> > | There are no IPv4 dependencies in this protocol.
> >
> > This specification will soon be officially obsoleted by
> > draft-ietf-hubmib-mau-mib-v3-04.txt, which will also
> > replace RFC 2668. Since draft-ietf-hubmib-mau-mib-v3-04.txt
> > is in the RFC Editor's queue as we speak, I think it would
> > be reasonable to drop RFC 1515 from the survey. Note that
> > it actually should have been obsoleted by RFC 2668 but was
> > not, owing to an oversight.
>
> (s/protocol/specification/ -- Andreas, please check that this is OK
> everywhere!)
>
> I suggest keeping it for now, but if you prefer, adding something like:
>
> Moreover, this document has been Obsoleted.
I don't like particularly like the idea, but I can live with keeping
it if you insist; since the replacement for RFC 2668 is not yet
published, this RFC 1515 is not yet "officially" obsoleted. If you
do keep the section, here is the rewrite I'd like to see:
5.018 RFC 1515 Definitions of Managed Objects for IEEE 802.3
Medium Attachment Units (MAUs)
This document is not considered in this discussion because it
should have been obsoleted by RFC 2668; due to an oversight,
however, it was not. That oversight will be corrected when
the replacement for RFC 2668 (now in the RFC Editor's queue)
is published.
> > | 5.020 RFC 1611 DNS Server MIB Extensions (DNS-S-MIB)
> > ...
> > | 5.021 RFC 1612 DNS Resolver MIB Extensions (DNS-R-MIB)
> > ...
> >
> > As Juergen Schoenwaelder has pointed out, RFCs 1611 and 1612
> > have been declared HISTORIC and should be dropped from the
> > survey. (Once a document has been declared HISTORIC, it is
> > no longer a standard of any kind.)
>
> Yeah, I would hope Phil would have known that when he made his analysis
> :-(
It wasn't Phil's error, since the reclassification from PS to
HISTORIC postdated his work. However ...
> I think we should keep these here (for consistency), but if you prefer,
> cut the unnecessary text out and just say something like:
I disagree, as all it does is waste space in the document and divert
the reader's attention. Unlike 1157, 1643, and 1515, we don't even
have the excuse that these things are widely deployed. They aren't:
see RFC 3197 Applicability Statement for DNS MIB Extensions, R. Austein,
November 2001, which says:
In view of the community's apparent total lack of interest in
deploying these MIB extensions, we recommend that RFCs 1611 and 1612
be reclassified as Historical documents.
In view of this I can see no reason to waste even one character of
text on these documents ...
> 5.021 RFC 1612 DNS Resolver MIB Extensions
>
> There are multiple IPv4 dependencies in this specification.
>
> (and not waste paper in going into details when we say in section 7 that
> it's historic..)
... but if you absolutely must keep them (which is not quite as
unjustified as 1441, which was simply an error from the beginning)
how about this:
5.020 RFC 1611 DNS Server MIB Extensions (DNS-S-MIB)
This specification has been reclassified as an HISTORIC
document and is not considered in this discussion.
5.021 RFC 1612 DNS Resolver MIB Extensions (DNS-R-MIB)
This specification has been reclassified as an HISTORIC
document and is not considered in this discussion.
and then removing the corresponding analysis in Section 7.
At least this will spare us an examination of the corpses.
> > | 5.029 RFC 1759 Printer MIB (Print-MIB)
> >
> > s/Print-MIB/Printer-MIB/
>
> remove ?
>
> > | There are no IPv4 dependencies in this specification.
> >
> > Note: this document is slated to be replaced someday "soon" by
> > <draft-ietf-printmib-mib-info-15.txt>. When you issue an update
> > of this draft, check the status of RFC 1759 to see if it has
> > been replaced.
>
> Ok, one could add, if you prefer, something like:
>
> Moreover, this specification is being Obsoleted.
I was just suggesting to check
ftp://ftp.isi.edu/in-notes/rfc-index.txt as the document was being
edited to see if the section should be re-titled with a new RFC
number. If you don't care to impose that burden on the editor I
would rather leave the section as-is. The moving target nature of
the subject matter makes some things like this unavoidable (even if
the drafts are up-to-date when approved, they'll have plenty of time
to grow stale while sitting in the RFC Editor's publication queue).
> > | 5.034 RFC 2020 IEEE 802.12 Interface MIB (802.12-MIB)
> >
> > s/802.12-MIB/DOT12-IF-MIB/
>
> remove ?
>
> > Note: the actual title is "Definitions of Managed Objects for
> > IEEE 802.12 Interfaces" ... I've not been checking this; does
> > it matter?
>
> we should use the correct title, I think.
So out would become (modulo a possible section number change)
5.034 RFC 2020 Definitions of Managed Objects for IEEE 802.12
Interfaces
There are no IPv4 dependencies in this specification.
OK by me.
> > | 5.058 RFC 2493 Textual Conventions for MIB Modules Using
> > | Performance History Based on 15 Minute Intervals
> > |
> > | There are no IPv4 dependencies in this specification.
> >
> > Note: this document is due to be replaced by RFC 3593,
> > which will be a Draft Standard and will have the same title.
That has now happened: the announcement has just been sent to the
IETF-Announce list, and RFC 3593 is now available in the on-line
repositories since yesterday. So, please change the RFC number from
2493 to 3593 and move this sub-section from the PS part to the DS
part. Yes, I know this contradicts your minimal change doctrine but
I don't agree with that doctrine.
> > | 5.066 RFC 2558 Definitions of Managed Objects for the SONET/SDH
> > | Interface Type
> > |
> > | There are no IPv4 dependencies in this protocol.
> >
> > s/protocol/specification/
> >
> > Note: this document is due to be replaced by RFC 3592,
> > which will be a Draft Standard and will have the title
> > "Definitions of Managed Objects for the Synchronous
> > Optical Network/Synchronous Digital Hierarchy (SONET/SDH)
> > Interface Type"
>
> If you prefer, we could add something like:
>
> Moreover, the document is obsoleted by RFC 3592.
>
> (or even move up to the draft standards if you really, really want)
I really, really, want, especially since RFC 3592 is now available
in the on-line repositories (cf. 2493 -> 3593 above). The relocated
section should just say:
4.xxx RFC 3592 Definitions of Managed Objects for the Synchronous
Optical Network/Synchronous Digital Hierarchy (SONET/SDH)
Interface Type
There are no IPv4 dependencies in this specification.
> > | 5.068 RFC 2562 Definitions of Protocol and Managed Objects for
> > | TN3270E Response Time Collection Using SMIv2 (TN3270E-RT-MIB)
> > | (TN2370E-RT)
> >
> > remove "(TN3270E-RT)" at the end
>
> remove both ? :-)
No, "(TN3270E-RT-MIB)" is actually part of the official title.
> > | Several OIDs rely on imports from RFC 2561 and therefore the
> > | protocol is both IPv4 and IPv6 aware.
> >
> > rewrite:
> >
> > This MIB module inherits IP version-independence by virtue of
> > importing the appropriate definitions from RFC 2561.
>
> ok
Now, since I promised specific text for documents that have been
updated since Phil's original survey:
> > | 5.070 RFC 2576 Coexistence between Version 1 Version 2 and Version
> > | 3 of the Internet-standard Network Management Framework (SNMP)
> >
> > remove "(SNMP)"
>
> ok for both
s/2576/3584/
Move to Section 3 as discussed above and assign a new sub-section number.
> > | This document states:
> > |
> > | (11) For any object with a SYNTAX of NetworkAddress, the SYNTAX MUST
> > | be changed to IpAddress. Note that the use of NetworkAddress in
> > | new MIB documents is strongly discouraged (in fact, new MIB
> > | documents should be written using SMIv2, which does not define
> > | NetworkAddress).
The equivalent text in RFC 3584 is:
(10) For any object with a SYNTAX of NetworkAddress, the SYNTAX MUST
be changed to IpAddress. Note that the use of NetworkAddress in
new MIB documents is strongly discouraged (in fact, new MIB
documents should be written using SMIv2, which does not define
NetworkAddress).
> > | and defines the OID:
> >
> > s/OID/object/
>
> ok
>
> > | snmpTrapAddress OBJECT-TYPE
> > | SYNTAX IpAddress
> > | MAX-ACCESS accessible-for-notify
> > | STATUS current
> > | DESCRIPTION
> > | "The value of the agent-addr field of a Trap PDU which
> > | is forwarded by a proxy forwarder application using
> > | an SNMP version other than SNMPv1. The value of this
> > | object SHOULD contain the value of the agent-addr field
> > | from the original Trap PDU as generated by an SNMPv1
> > | agent."
> > | ::= { snmpCommunityMIBObjects 3 }
This is unchanged in RFC 3584.
> > | This clearly points out a lack of IPv6 awareness in this specification.
> >
> > That last sentence is incorrect. The presence of this stuff
> > is not because the specification has a lack of IPv6 awareness;
> > it is there because it is necessary to allow a MIB translator
> > or a proxy forwarder to do its job ... and that job is to allow
> > legacy versions of the SNMP (SNMPv1 and SNMPv2c, both now Historic
> > protocols) to coexist with SNMPv3. I think Randy Presuhn has
> > complained about this before. Here is the suggested replacement:
> >
> > Since purpose of this document is to describe how legacy
> > specifications with IPv4 dependencies (SMIv1, SNMPv1, and
> > SNMPv2c) can coexist with current ones (SMIv2 and SNMPv3),
> > the IPv4 dependencies documented above are acceptable and,
> > indeed, unavoidable.
> >
> > In addition, note that RFC 2576 has been replaced by RFC 3584,
> > which is a BCP. So the title needs to changed, and sub-section
> > itself needs to be moved.
>
> ok
Glad you agree, details filled in above.
> > | 5.083 RFC 2665 Definitions of Managed Objects for the
> > | Ethernet-like Interface Types (MIB)
> >
> > s/MIB/EtherLike-MIB/
>
> remove?
>
> > | There are no IPv4 dependencies in this specification.
> >
> > Note: this document is due to be replaced by
> > draft-ietf-hubmib-etherif-mib-v3-03.txt, which
> > the IESG recently approved as a Proposed Standard.
> > When you issue an update of this draft, check the
> > status of RFC 2665 to see if it has been replaced.
>
> ok, could add something like, if you prefer:
>
> Moreover, this specification is being Obsoleted.
I'd rather not add that text; it would probably be better to just leave
it alone. Please see the note under RFC 1759 for more explanation.
> > | 5.085 RFC 2668 Definitions of Managed Objects for IEEE 802.3 Medium
> > | Attachment Units (MAUs) (MAU-MIB)
> > |
> > | There are no IPv4 dependencies in this specification.
> >
> > Note: this document is due to be replaced by
> > draft-ietf-hubmib-mau-mib-v3-04.txt, which the
> > IESG recently approved as a Proposed Standard.
> > When you issue an update of this draft, check the
> > status of RFC 2668 to see if it has been replaced.
>
> like above
yes, like above :-)
> > | 5.092 RFC 2726 PGP Authentication for RIPE Database Updates
> > |
> > | There are no IPv4 dependencies in this protocol.
> >
> > Is this actually an OPS Area specification?
>
> this could be moved to security, that's true. however, we felt that RPSL
> was so strongly OPS stuff, so this could logically belong here too
> (doesn't really matter as there are no dependencies).
I was just asking. If you think it's in the right place leave it
(RPSL evidently is, according to Bert Wijnen).
> > | 5.093 RFC 2737 Entity MIB (Version 2)
...
> > [ ... ] I therefore propose to replace the preceding
> > two paragraphs with this:
> >
> > There are no IPv4 dependencies in this specification.
>
> perhaps this could be spelled out a bit more, but ok.
>
> > Note: this RFC is due to be replaced by draft-ietf-entmib-v3-03.txt
> > or a successor; when published it will be a Draft Standard. Check
> > ftp://ftp.isi.edu/in-notes/rfc-index.txt before updating this memo
> > to determine if this has happened.
>
> would you prefer to add a piece of text on that?
No, see note under RFC 1759.
> > | 5.098 RFC 2769 Routing Policy System Replication (RPSL)
> >
> > remove "(RPSL)"
>
> agree!
>
> >
> > | There are no IPv4 dependencies in this protocol.
> >
> > Is this an OPS Area spec? It looks like a routine area
> > spec to me.
>
> routing area, yes, could be a target. but all of the RPSL stuff is in
> here, so perhaps it's best to keep it here esp. because there are no
> dependencies. But ok to move if you prefer.
Bert Wijnen agrees that it belongs here (I was just asking, since it
seemed "suspicious").
> > | 5.102 RFC 2837 Definitions of Managed Objects for the Fabric Element
> > | in Fibre Channel Standard
> > |
> > | There are no IPv4 dependencies in this protocol.
> >
> > s/protocol/specification/
>
> ok (I still can't figure out why when the editor changed some of those to
> "specification" he didn't change all.. but have been a bug or something..)
>
> > | 5.103 RFC 2851 Textual Conventions for Internet Network Addresses
> > |
> > | This MIB defines a new set of OIDs for that allow new MIB's to
> > | use multiple versions of IP. Currently IPv4 and IPv6 addressing
> > | is defined. Update of the many MIBs previously identified as
> > | having IPv4 dependencies could easily be updated using this new
> > | set of IP address abstractions.
s/OIDs/TCs/
or if you prefer
s/OIDs/textual conventions/
> > This document has been replaced by RFC 3291 (also a proposed
> > standard). Please update this subsection accordingly.
>
> does it change the content of the issue, described above?
>
> ok to change.
Nothing needs to be changed except the RFC number. You could use
the text as-is.
> > | 5.113 RFC 2959 Real-Time Transport Protocol Management
> > | Information Base
> > |
> > | There are numerous uses of the included TAddress Syntax which is
> > | IPv4 dependent as noted above.
> > |
> > | For example:
...
> > | There are a total of 8 instances of this.
> >
> > As explained above under 5.093, this does not necessarily mean that
> > there is an IPv4 dependency. Please re-evaluate this MIB module and
> > update this section accordingly. I'd bet that this spec doesn't really
> > have any IPv4 dependencies after all.
>
> we don't have resources for re-evaluation, unfortunately. unless someone
> more knowledgeable does it, we'll have to keep it as is.
OK, I just had a quick look at this RFC and it looks OK to me ... the
object rtpSessionDomain, which is of type TDomain, allows IPv6 transport
to be represented by all the TAddress objects. So here's the rewrite:
5.113 RFC 2959 Real-Time Transport Protocol Management
Information Base
There are no IPv4 dependencies in this specification.
(modulo a probable sub-section number change).
> > | 5.121 RFC 3084 COPS Usage for Policy Provisioning (COPS-PR)
> > | (COPS-PR)
> >
> > remove redundant "(COPS-PR)"
> >
> > | This is an IPv4 only protocol. A version for IPv6 must be defined.
> >
> > s/must be/may ned to be/
try that again:
s/must be/may need to be/
Please consider adding a sub-section for this brand-new proposed standard:
5.xxx RFC 3591 Definitions of Managed Objects for the Optical
Interface Type
There are no IPv4 dependencies in this specification.
... skipping down to the experimental RFCs ...
> > | 6.06 RFC 1901 Introduction to Community-based SNMPv2 (SNMPV2CB)
> > |
> > | There are no IPv4 dependencies in this protocol.
> >
> > RFC 1901 is now HISTORIC and should be dropped from this survey.
>
> should keep it here, but could add a statement here, like:
>
> Moreover, this specification has been reclassified HISTORIC.
If you insist on keeping it, here's the appropriate rewrite:
6.06 RFC 1901 Introduction to Community-based SNMPv2 (SNMPv2c)
There are no IPv4 dependencies in this protocol. (Note
that it has been reclassified as HISTORIC but is still
widely deployed.)
In case you keep this leave "(SNMPv2c)" in the section heading, since
it's probably better known under that name than any other.
> > | 6.07 RFC 1909 An Administrative Infrastructure for SNMPv2
> > | (SNMPV2AI)
> > |
> > | There are no IPv4 dependencies in this protocol.
> >
> > RFC 1909 is now HISTORIC and should be dropped from this survey.
> >
> > | 6.08 RFC 1910 User-based Security Model for SNMPv2 (SNMPV2SM)
> > |
> > | There are no IPv4 dependencies in this protocol.
> >
> > RFC 1910 is now HISTORIC and should be dropped from this survey.
>
> same for both of these
These are not now and never were very widely deployed, so keeping
them in the survey does not IMnsHO serve a useful purpose. But,
if you do insist on keeping them around, here are the rewrites:
6.07 RFC 1909 An Administrative Infrastructure for SNMPv2
This specification has been reclassified as an HISTORIC
document and is not considered in this discussion.
6.08 RFC 1910 User-based Security Model for SNMPv2 (SNMPV2SM)
This specification has been reclassified as an HISTORIC
document and is not considered in this discussion.
... skipping down to the analysis section ...
> > | 7.1 Standards
> > |
> > | 7.1.1 STD 15 Simple Network Management Protocol (RFCs 1157, 1155, 1213)
> > |
> > | The limitations identified have been addressed; RFC 1157 is HISTORIC,
> > | RFC1155 is obsoleted by RFC 2578-2580, and RFC1213 has been split to
> > | multiple modules which have been seen to.
> >
> > The above actually concerns three separate STDs.
> >
> > STD 15, RFC 1157, is SNMPv1, but it has been retired ... that is, it has
> > been reclassified as HISTORIC and therefore is no longer an Internet
> > Standard. So it should be dropped from the survey.
>
> disagree, due to the process differences (if we just had the time
> machine..)
As I said above, I agree that there is some practical justification
for keeping RFC 1157 in the survey since it is still widely deployed,
and I can live with your proposed rewrite, modulo a little
wordsmithing. Here goes:
7.1.1 STD 15 Simple Network Management Protocol (RFC 1157)
The limitations identified have been addressed: RFC 1157 has
been reclassified as HISTORIC, and the replacement documents
(RFCs 3411-3415) are compatible with both IPv4 and IPv6.
7.1.2 STD 16, Structure of Management Information (RFCs 1155 and 1212)
> > RFCs 1155 and RFCs 1212 (along with the informational document RFC
> > 1215) define SMIv1. These documents have been superseded by RFCs
> > 2578, 2579, and 2580 which define SMIv2. Since SMIv1 is no longer
> > being used as the basis for new IETF MIB modules, the limitations
> > identified in this Internet Standard do not require any action.
>
> ok.
7.1.3 STD 17, Management Information Base MIB II (RFC 1213)
> > The limitations identified are being addressed: the IP group,
> > the ICMP group, the TCP group, and the UDP group are being
> > replaced by IP version-independent MIB modules now in development
> > in the IPv6 working group (for more details see the entries
> > below for RFCs 2011, 2012, 2013, and 2096). No replacement is
> > being contemplated for the EGP group because EGP is a historic
> > protocol that is no longer in significant use in the Internet.
>
> ok.
... skipping down to the draft standards ...
> > | 7.2.3 RIPv2 MIB (RFC 1724)
> > |
> > | See Section 7.1.24. This problem is currently being addressed by the
> > | RIP WG and an ID exists (draft-ietf-rip-mib-01.txt).
> >
> > Where is Section 7.1.24?
>
> routing, now..
So update the cross-reference, or remove it.
> > I agree with the 2nd sentence, except
> > that the draft is now draft-ietf-rip-mib-01.txt; it stands alone
> > quite nicely.
>
> the draft name you mentioned is the same as listed, perhaps you meant
> something else? (btw, the draft has expired in the meantime)
I think I meant to say draft-ietf-rip-mib-02.txt but as you say that's
only a note saying that the draft has expired and the WG has concluded.
So how about this:
7.2.3 RIPv2 MIB (RFC 1724)
There is no updated MIB module to cover the problems outlined. A
new MIB module should be defined.
... skipping down to the proposed standards ...
> > | 7.3.03 DNS Server MIB (RFC 1611)
> > |
> > | This document is HISTORIC and no action is required.
> >
> > If it's HISTORIC it's not a Proposed Standard anymore.
> > Drop it from the survey.
>
> disagree
Even if you leave it in the survey there is no point
in analyzing it, especially if you update the stuff
in section 5.020 as I suggested (saying that the
spec has "been reclassified as an HISTORIC document
and is not considered in this discussion").
> > | 7.3.04 DNS Resolver MIB (RFC 1612)
> > |
> > | This document is HISTORIC and no action is required.
> >
> > If it's HISTORIC it's not a Proposed Standard anymore.
> > Drop it from the survey.
>
> see above
s/5.020/5.021/ in my reply
> > | 7.3.05 Appletalk MIB (RFC 1742)
> > |
> > | The problems have not been addressed and a new MIB should be defined.
> >
> > rewrite:
> >
> > This problem has not been addressed. If a user requirement for
> > IPv6 over Appletalk develops (which is thought to be unlikely)
> > then this MIB module will need to be updated (or a new MIB module
> > will need to be created) in order to accomodate it.
>
> I'd soften the tone a bit, because AFAIK, some Apples are still using
> appletalk, so
>
> s/is thought to be unlikely/may be unlikely/
>
> but ok as is too.
softening the tone is OK by me
> > | 7.3.10 RMON MIB (RFC 2021)?
> >
> > s/RMON MIB/RMON-II MIB/
> > remove question mark
> >
> > | The problems have been addressed in RFC 2819, Remote Network
> > | Monitoring Management Information Base.
> >
> > That is false: RFC 2021 is RMON-II, and RFC 2819 is the full
> > standard version of RMON-I (formerly RFCs 1271 and 1757). There
> > is a work underway to update RFC 2021, however; so the correct
> > statement is probably this:
> >
> > This issue is being resolved by the RMONMIB WG and there is an
> > ID (draft-ietf-rmonmib-rmon2-v2-00.txt).
> >
> > In fact draft-ietf-rmonmib-rmon2-v2-00.txt doesn't actually address
> > the problem but I have raised the issue on the RMONMIB mailing list.
>
> so, maybe reword:
>
> This issue is being resolved by the RMONMIB WG and there is an
> ID (draft-ietf-rmonmib-rmon2-v2-00.txt); future revistions may
> address the problem as it has been identifier and brough to attention.
The reworded text itself needs rewording :-) How about:
This issue has been brought to the attention of the RMONMIB WG.
Currently there is an ID (draft-ietf-rmonmib-rmon2-v2-00.txt)
to update RFC 2021, but it does not address the problems that
have been identified; it is expected that there will be a
resolution in a future version of that ID.
For those who are curious, look here:
http://www1.ietf.org/mail-archive/working-groups/rmonmib/current/maillist.html
to see the discussions over the past few days about this. One
possibility is to and an enum to trapDestProtocol and to update
the trapDestAddress DESCRIPTION clause. Another is to do nothing,
and insert text rfc2021bis saying that the tables in RFC 3413
should be used to provide this functionality. Stay tuned!
> > | 7.3.11 DataLink Switching using SMIv2 MIB (RFC 2022)
> >
> > s/2022/2024/
> > Note: RFC 2022 is the multicast over ATM spec (MARS and so on)
> >
> > | The problems have not been addressed and a new MIB should be
> > | defined.
> >
> > s/a new/an updated/
>
> ok
>
> > | 7.3.12 IP Forwarding Table MIB (RFC 2096)
> > |
> > | This issue is being worked on by the IPv6 WG and an ID exists to
> > | address this (draft-ietf-ipngwg-rfc2096-update-00.txt)
> >
> > s/draft-ietf-ipngwg-rfc2096-update-00.txt/draft-ietf-ipv6-rfc2096-update-05.txt/
> > (it was posted on 29 Aug 2003, should be listed on IPv6 WG charter page by now)
>
> ok
>
> > | 7.3.13 Classical IP & ARP over ATM MIB (RFC 2320)
> > |
> > | The problems identified are not addressed and a new MIB must be
> > | defined.
> >
> > The previous paragraph should be rewritten along the following
> > lines:
> >
> > The current version of Classical IP and ARP over ATM (RFC xxxx)
> > does not support IPv6. If and when that protocol specification
> > is updated to add IPv6 support, then new MIB objects to represent
> > IPv6 addresses will need to be added to this MIB module.
>
> ok, replace xxxx in to point to the CLIP RFC though (in routing survey?)
Oops, I think I meant to fill that in and forgot. It's RFC 2225.
> > | 7.3.14 Multicast over UNI 3.0/3.1 ATM MIB (RFC 2417)
> > |
> > | The problems identified are not addressed and a new MIB must be
> > | defined.
> >
> > The previous paragraph should be rewritten along the following
> > lines:
> >
> > The current version of Multicast over UNI 3.0/3.1 ATM (RFC
> > xxxx) does not support IPv6. If and when that protocol
> > specification is updated to add IPv6 support, then new MIB
> > objects to represent IPv6 addresses will need to be added to
> > this MIB module.
>
> ok, same here.
RFC 2022. Sorry about that.
> > | 7.3.18 Coexistence of SNMP v1, v2, & v3 (RFC 2576)
> > |
> > | There are no real issues that can be resolved.
> >
> > rewrite:
> >
> > There are no issues that need to be resolved.
> >
> > Note also that the coex doc is now RF 3584, which is a BCP.
>
> ok
Just to be clear, the last sentence an instruction to the editor,
not text to be included. That is, the updated text (module sub-
section number) would read:
7.3.18 Coexistence of SNMP v1, v2, & v3 (RFC 3584)
There are no issues that need to be resolved.
> > | 7.3.31 RPSL (RFC 2622)
> > |
> > | Additional objects must be defined for IPv6 addresses and prefixes.
> >
> > s/must/should/
>
> Keep it as MUST.
As I understand the V6OPS WG charter, an update to the RPSL MIB (or
to RPSL itself) is out of scope for this WG. If that understanding
is correct, a V6OPS document can't mandate an uppercase MUST here.
Anyway, there are other things wrong with this text since the spec
defines a language, not a MIB module (as suggested by the word "objects").
> but you can say:
>
> draft-blunk-rpslng-01.txt defines extensions to solve this issue, and it
> is being considered for publication.
How about this:
7.3.31 RPSL (RFC 2622)
The ID draft-blunk-rpslng-01.txt defines extensions for IPv6
addresses and prefixes, and and it is being considered for
publication.
We agree on the fixes to the analysis of experimental RFCs,
so we're done.
> THANKS! We haven't had such an extensive review in ages... :-)
Thank you for yout detailed reply. I hope that I have strong-armed
you into making a few more changes than your "minimum update" policy
would mandate, maybe with the help of the carrot of providing some
actual text :-)
//cmh