Joe:
Excellent questions. The intent was to spin down the *hares* drafts. I wished you’d asked the question in the meeting. In my experience, you questions are always timely. Please start a YANG module draft for I2RS. My recommendation to I2RS/NETCONF/RESTCONF chairs was: · Expand revised datastores – to allow dynamic datastores + name · Add capabilities to NETCONF/RESTCONF for dynamic datastores and capability · Get volunteers from I2RS/NETMOD work on YANG syntax in NETMOD · Get volunteers from I2RS/NETCONF work on the NETCONF changes in NETCONF · Get volunteers from I2RS/NETCONF work on the RESTCONF changes for ephemeral in NETCONF. Open issues for these people: · Precedence: (yang/protocols) – must determine how to prioritize between datastores during installation of configuration · Validation (yang/protocols) - does it need to be identified in YANG or is it an implementation specific? · RESTCONF (protocol) - how to encode client-id/priority in entity-tag · NETCONF (protocol) – how to handle edit-collision (aka multiple client write) · Protocol support – how do you note what protocol and protocol features are needed to support this model. Now -- to your specific comments: · Draft: This draft was intended to be a chair's kick start document on open issus. · Why Kickstart: I started it when the revised datastores people did not open their github archive or publish drafts to make sure we covered this work at IETF. o Fortunately – they did release -01.txt revised datastores. · Content errors: o For edit-config, see the comments by Phil Schafer. o The ephemeral logic is backward = (red face) o Revised datastores revisions-01 – did the following old/control-plane datastores/dynamic datastores/ (my draft worked off -00.txt) Let me know if I’ve answered all the questions. Sue Hares -----Original Message----- From: i2rs [mailto:[email protected]] On Behalf Of Joe Clarke Sent: Wednesday, March 29, 2017 4:09 PM To: [email protected] Subject: [i2rs] Comments on I2RS and YANG suggestions (aka, finally woke up) Lunch has been really good here at the Swissôtel, but that wasn't the reason I didn't get up at the mic. I have been going through the various YANG and *CONF drafts from Sue this week, and I was hoping the presentation would help in my clue on some things. After mulling things over, it did not. At the risk of asking stupid questions, I will comment. First, there appears to be a big typo in draft-hares-netmod-i2rs-yang-04 that may change my understanding of how to specify something is ephemeral. If you look at section 3.4, it states: The value "true" indicates the object is not ephemeral. The value "false" indicates the value is ephemeral. I think (hope) that is backwards. But if not, can you explain to me why the logic is thus? === Now on to my bigger questions/comments. I have been trying to figure out why we need a dstype. I see that the examples listed are control-plane, config, and i2rs-v[o0] (this shows up in various spellings without any explanation). I see no mention of the idea of a "control-plane" DS in the revised DS draft, and I don't see why we need this added complexity. The revised DS draft already allows us to specify a DS name, additional attributes such as "ephemeral," protocol support, and supporting modules. Why add this DS type? What does it give us and how would one figure out how to make use of it? I'm strongly leaning on letting this draft expire and working to create an I2RS YANG module that fits into the revised DS guidance. I think that would make things much clearer. === Additionally, if the revised DS draft progresses (and it seems like it will), do we need <write-data> at all? Yes, we need to sort out validation, but why can't we use <edit-config> (revised DS already gives us <get-data>). === Again, sorry if these are dumb questions. It seems like the work around revised datastores is already addressing most of what we need. If the WG agrees, I think we should focus on removing confusion and ambiguity and work to define clearly distinct documents that can inform the current world order. Thanks. Joe _______________________________________________ i2rs mailing list <mailto:[email protected]> [email protected] <https://www.ietf.org/mailman/listinfo/i2rs> https://www.ietf.org/mailman/listinfo/i2rs
_______________________________________________ i2rs mailing list [email protected] https://www.ietf.org/mailman/listinfo/i2rs
