Carl Lebsack

The request should be a named request in your config file as it gets added to a Tunnel target dynamically. The tunnel request commandline gets added to a tunnel target...

On Wed, Mar 8, 2017 at 1:56 AM, Benoit Sigoure wrote: > gNMI currently only supports dial-in (i.e. a client coming from the > outside and connecting to the network...

I would counter propose a dial-out service that has exactly the same semantics as the original Subscribe, only the direction of the establishing of the connection changes. Meaning that a...

On Wed, Mar 8, 2017 at 12:30 PM, pborman wrote: > The question becomes, in a dial-out service, who is responsible for > determining what data is sent. It seems...

grpctunnel is the approach that has been proposed and discussed with multiple vendors and network operators as a general solution that addresses many different use cases, including dial-out telemetry. For...

The gNMI specification is unchanged for the proposed dialout mechanism using grpctunnel. The device dials a tunnel session to a collector which is used as a trigger to make a...

The diagram matches what the current specification dictates if updates_only is set to true, with the addition that steps 2-4 can be repeated any number of times on the same...

I concur with @robshakir here. The single timestamp was a deliberate decision given the streaming implementation and the goal for low-latency updates. Data that is collected together can be sent...

@samante This sounds like a misattribution of timestamps, not the need to send data with different timestamps in a single payload. If a line card is sending a payload every...

@samante The current encoding allows for correct timestamps to be assigned to every given statistic. gNMI is a stream and thus we are talking about whether to send multiple notifications,...