Skip to content

Minutes 2016 10 05

Jump to bottom
ericvoit edited this page Oct 5, 2016 · 2 revisions
Meeting Materials Attending
WebEx Recording password: TwhNuRi4 Andy Bierman, Alexander Clemm, Alberto Gonzalez Prieto, Eric Voit, Susan Hares, Tim Jenkins, Balazs Lengyel, Kent Watsen, Michael Scharf

Highlights

  • Anydata instead of string for yang push updates.

  • We will use Anydata instead of string for supporting encoded push data. Not just JSON or XML. (Protobuf / CBOR is easily carried too with Anydata.)

  • Annotation proposal

  • Open question is whether we have a new (tbd) statement to allow non-notifiable and other assertions to be made against leafs. This generalized assertion capability is not really yang-push specific.

  • Balazs to go the Netmod and Netconf alias to socialize proposal to see if a general statement flies. If WGs don't like a generalized capability, we could have something just for the not-notifiable capability for yang-psuh.

  • Augmentations must be allowed outside of base yang file to improve version control.

  • Resetting on-change subscription state

  • Ability to send a push-update (rather than a push-change-update) to reset an existing on-change subscription to a clean state. (This may or may not be synchronized with a suspend/resume.)

  • Was a request to have an RPC for dynamic subscriptions to allow the client to request a resynch.

  • Anchor time

  • We reaffirmed we we need anchor-time in addition to start-time from 5277bis.

  • Adding “Notifications-not-sent” flag to push-update notifications.

  • Indicates that objects were not included based on an inability of the publisher to fulfill the contract.

  • It is possible to send such an update with object contents.

  • Is relevant for on-change and periodic update.

  • Will not use independent notification to allow easy synchronization with contents and periods of specific push updates

  • Simplifying Modify subscription parameters

  • Removed the following from yang-push modify-subscription: subscription-priority, subscription-dependency, dscp. Need to also removing synch-on-start and excluded change.

  • We should also remove stream and encoding for 5277bis.

  • Etag

  • At this point we will not pursue for yang-push. Difficulties include etag on identifies on the wire contents, doesn't send etags for each of the pushed objects.

  • Unencrypted HTTP2 transport

  • As HTTP2 doesn't require encryption, this should not be an issue.

Next Meeting

  • Review new update to yang-push draft and model. If time permits, work on stream definitions.
  • Get ready for IETF S Korea
  • Do we want an informal session on transport of GBP/CBOR yang encodings?
Clone this wiki locally