So in the flume architecture we have data send to a source, the source write to the channel and the channel send the event to the sink that does what it needs to do.
I currently have a flume agent using a memory channel, a AvroSource and and a HdfsSink.
To send data to the AvroSource, I use the flume-ng-sdk that propose me the appendBatch method.
This appendBatch method throw an exception if the event isn't delivered to "at least his next hop" as the documentation says.
An event delivery exception is raised whenever an Event fails to reach at least one of its intended (next-hop) destinations.
But my question here is, what is the "intended next-hop destination" here ? Is it the channel ? Or is it the source ?
I think it's very important and I struggle to understand it well.
Some use case needs to ensure that the event was received on the channel because it's on the channel that we have some persistence feature.