draft-ietf-vrrp-unified-mib-07.txt MIB Dr. Review

Hello Kalyan,

The examples in the document are very helpful and
there was a great deal of work that went into this document.
Thank you for that!

MIB compiler output is first, followed by comments.

Thanks,
  -Joan


MIB Compiler: SMICng
--------------------

W: f(VRRP-MIB.my), (437,8) Row "vrrpAssociatedIpAddrEntry" has indexing
   that may create variables with more than 128 sub-ids
W: f(VRRP-MIB.my), (158,23) Row "vrrpOperationsEntry" does not have a
   consistent indexing scheme - index items from current
   table must come after index items from other tables
W: f(VRRP-MIB.my), (453,23) Row "vrrpAssociatedIpAddrEntry" does not have
   a consistent indexing scheme - cannot specify an index item from
   additional "base row" vrrpOperationsEntry, since can have
   only one "base row" which is ifEntry


WRT the INDEX To Large error.  I think this was addressed by Bert Wijnen in 
prior
versions but I don't see any wording in the DESCRIPTION clause or any SIZE 
limit
on the Index vrrpAssociatedIpAddr.  So something like this (and some text 
describing
why the size limits in the DESCRIPTION clause would be appropriate. 
Additionally,
the vrrpOperationsInetAddrType should be restricted.


       vrrpAssociatedIpAddr OBJECT-TYPE
           SYNTAX       InetAddress (SIZE(0|4|16))
           MAX-ACCESS   not-accessible
           STATUS       current
           DESCRIPTION


Please discuss the other 2 warnings wrt Indexing.  Specifically, why are
tables using AddrType, VrId and ifIndex, rather than ifIndex, VrId, 
AddrType?
(Please see comment below regarding section 8.)



MIB Compiler: SmiLint
------------------------
mibs/VRRP-MIB.my:437: [5] {index-exceeds-too-large}
   warning: index of row `vrrpAssociatedIpAddrEntry'
   can exceed OID size limit by 142 subidentifier(s)


Comments
--------

1)NIT:  Does not strip cleanly with SmicngPRO mstrip
(This doesn't need to be fixed for MIB Dr. review, just
as an fyi).

2) Please Discuss:
My understanding (based on an old email exchange (fall 2008) with the MIB 
author
and the AD of the WG at that time) was that the VRRP unified MIB
was not going to be an update of the original VRRP-MIB (RFC2787), but
rather a new MIB based on the new unified spec that was being
created.  There are a couple issues that I'd like clarified wrt to an
updated MIB (this draft) vs. a new MIB:


A) draft-ietf-vrrp-unified-spec-05.txt, section 8.4 states "that VRRPv2 and 
VRRPv3
interoperation is optional" and the
recommendation is that it should only be done for transitioning from
VRRPv2 to VRRPv3.  A new MIB (rather than an updated
MIB) would likely re-enforce the guidelines from the specification.
Could you discuss why an updated MIB is better in this regard?


B)  Additionally, this new MIB is based on a MIB that is 10 years old 
(RFC2787),
while not specifically an issue, I do see that this updated MIB is
using older MIB conventions.  Not a show-stopper but certainly a
consideration.  Was this considered?

C) Comment: Most of the MIB objects contained in RFC2787 are deprecated
by this MIB. This indicates that creating a new MIB would be
fairly straightforward.


3) The terminology of this document does not consistently utilize the 
Definitions
outlined in the draft-ietf-vrrp-unified-spec-05.txt.  So for example,
VRRP is used in places to mean VRRP Router.  The use of IPvX is not
done.  Would be better to be consistent between the spec and the MIB.



4)  DEPRECATED OBJECT DESCRIPTIONS:

Way back when, Dave Perkins suggested starting the DESCRIPTION clause
for a deprecated object with:

"This object is DEPRECATED.  This is done because..."
followed by the original description.

While you have done a great job including the reasons why an object was
deprecated, could you move this to the beginning of the DESCRIPTION clause
(rather than end)?



5) Title: Please mention that this is Version 3 somewhere in the title
(as does the specification)


6) Section 4. Relationship to RFC2787

The following sentence is troublesome because the MIB Module "deprecates"
objects and does not obsolete them.  While you may be trying to indicate
that this document will obsolete RFC2787, that is not exactly clear.

(perhaps, this is another reason to create a new MIB module vs. an updated
one?)

   "RFC2787 [RFC2787] defines managed objects for VRRP over IPv4 and is
   now obsoleted by this memo."

7) Section 5. Relation to Interface Group (IF-MIB)

A) Please remove the word "physical" in front of physical interface.

B) Should specify a reference such as:  The VRRP-MIB Module imports
ifIndex from the IF-MIB.  At this time, the latest version
of IF-MIB is from [RFC2863].


C) A reference to [RFC2863] needs to be provided in the
Normative References.


8) Section 7. VRRP MIB Structure


This section should discuss the relationship between the tables.

The groups listed are part of the conformance, and while okay
to mention these, that really isn't the MIB's structure.

NIT: Should have a (3) before "The vrrpAssociatedIpAddrTable"


9) Section 8. VRRP MIB Table Design

The MIB design should focus on indexing the tables such that
the table lookups are more advantageous for the device (e.g. router),
 not the NM app.  The NM app should be able to
handle rearranging data for display purposes.

I would like to see a discussion of the indexing wrt the device.
The warnings from the MIB compiler indicate that there may be a
better indexing (table structure) available.  That doesn't mean
that you need to resolve the warnings, but I would like to see
some text included which shows why the chosen indexes are
advantageous for the VRRP router, and the Virtual Routers.

IMHO, if an interface goes down,
then would be probably more advantageous
to know the Virtual router(s) associated with that interface.
(and subsequently the IP addresses on those Virtual Routers).

Please combine Section 7 and Section 8.


10) Section 9.

Thank you for these examples.

11) Section 10.  Please specify that this is
the VRRP MIB Module Definition

* DATE and REVISION dates need to be updated.

* VrId (Textual Convention)

There is already a VrId in RFC2787. While it looks as if the
only difference is the DESCRIPTION clause, need to caution against
doing redefining this.  I would suggest creating a VRID TC and
making the DESCRIPTION simple, for example:  This value uniquely identifies 
a
Virtual Router on a VRRP router.

While the remaining text is informative, it is not really necessary.
Please also give a REFERENCE clause for this.


* vrrpNodeVersion

This scalar has correctly been deprecated, but I don't see any
version object replacement in a table.  Why not?  (There is
field in the protocol for this.)


* vrrpNotificationCntl, vrrpTrapNewMasterCntl, vrrpTrapProtoErrorCntl

Could notifications be generated or not, using
RFC3413?


*StorageType objects:

If an operator configures the VRRP Router with Virtual Router(s)
and IPvX addresses then I suspect that this information would be
saved to nonVolatile storage.  Do you agree?

In other words, I'm trying to think of a scenario when an operator
wouldn't want to save this information (once he/she is satisfied with
the configuration) in the sake of simplification, would there be
benefit to just state that the configuration should be saved
to NV storage?  If so, then
the use of StorageType objects in this MIB could be removed and
replaced with text in the Table Entry DESCRIPTION.

(RFC4181, Section 4.6.4 specifies that StorageType objects can be
used OR, the Table Entry DESCRIPTION clause specify what happens
to dynamically-created rows after an agent restart.)


* RowStatus Objects

There is some discussion about having to have the corresponding
Virtual Router in vrrpOperationsState have the value of
'initialize' before changing a value in the Row, and then the
RowStatus object is supposed to not be "active(1)", but I don't
understand what values are acceptable for modifying a read-creat
object in a row?  Also, how does this impact vrrpOperationsState?

* NIT: AdminState and OperationsState, could these be renamed to use
use AdminStatus and OperStatus?


* vrrpRouterVrIdErrors

Please verify that the DESCRIPTION is still valid for this MIB?
Should a value be added to the Statistics Table instead (i.e. on
a per Virtual Router basis?


* Please refrain from using the term "trap" in the current 
objects/notifications.
Please use the term "notification" as appropriate for the
current objects/notifications.  (Deprecated items are fine to leave the term
trap.)


* vrrpNewMasterReason, vrrpTrapProtoErrReason

Where are these Enum values from?  Could you add a reference?
Also, please explain the values.


* Conformance looks fine, but will review again once above MIB
Module comments are addressed.  Same with security section.

Thanks,
  -Joan



_______________________________________________
vrrp mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/vrrp