问题
RFC 5277 defines notification replay support. Just wondering what customer problems this notification replay can solve? What could be the need to scan through list of past notifications? At any point of time, controllers can always fire "get" RPC and determine current state.
Any inputs on this are appreciated.
回答1:
Notification replay is mostly required to sync NETCONF clients with servers without the need to do a full RPC. In some systems, the configuration + operational data becomes quite large, so the notification replay allows for a delta sync.
The main problem with this is that the server will only keep a limited time window of notifications; so if a client asks for a replay of the last week, the server may be only able to provide a subset of that.
RESTCONF has a different solution for this same issue, the last-modified header field(https://tools.ietf.org/html/rfc8040#section-3.4.1.1), which allows for doing a get of 'data that has changed since timestamp x'.
来源:https://stackoverflow.com/questions/57198426/netconf-notifications