On Thu, Nov 9, 2017 at 8:59 AM, Balazs Lengyel <balazs.leng...@ericsson.com> wrote:
> Hello, > In draft-ietf-netmod-rfc7223bis-00 I read > > The "/interfaces-state" subtree with "config false" data nodes is > deprecated. ... > > Servers that do not implement NMDA, or that wish to support clients > that do not implement NMDA, MAY implement the deprecated > "/interfaces-state" tree. > > This means that a server MAY simply remove a big branch > (interfaces-state). If I have a network management system (NMS) and the > YANG server upgrades to a new revision of the module suddenly my NMS SW > breaks. > > I find it really strange that, on a one hand YANG has very strict rules > about updating a module, on the other hand just by deprecating something > you can remove anything. Are we taking backward compatibility seriously or > not? It seems not. > > Ironically, the NMDA approach was sold as the solution path without any disruption to the data models. Customers will pressure vendors to fix "stuff that used to work" and generally vendors avoid breaking it in the first place. 3rd party NMS has never been a priority anywhere, including the IETF. Earlier we had discussions about the definition of deprecated and obsolete > in https://tools.ietf.org/html/rfc7950#section-7.21.2. > From an NMS point of view the current definition just states: Anything can > be deprecated/obsoleted anytime. If something has a status different from > current there is no guarantee its usable or even that it exists. > > IMHO these definitions are unusable for an NMS. Ericsson needed to > introduce its own rules. IETF should do something about this! I have > proposals if there are people interested. > > As a minimum I would propose, that a server that does not implements a > fully functional /interfaces-state" subtree MUST obsolete it, not just > deprecate it. > how does this help? Duplicate read-only data does not break anything. An old client can keep reading /interfaces-state and a new client can use <get-data>. There are some problems with NMDA that break old clients (e.g., rpc and action) but this isn't one of them. > regards Balazs Andy > > > -- > Balazs Lengyel Ericsson Hungary Ltd. > Senior Specialist > Mobile: +36-70-330-7909 email: balazs.leng...@ericsson.com > > _______________________________________________ > netmod mailing list > netmod@ietf.org > https://www.ietf.org/mailman/listinfo/netmod >
_______________________________________________ netmod mailing list netmod@ietf.org https://www.ietf.org/mailman/listinfo/netmod
