Problem

The receiver shuts down immediately upon receiving a sigterm and since we
close the HTTP server immediately and the endpoints of the service isn't
yet fully propagated in the cluster there might be requests dropped.

To make the receiver more resilient we need to wait x seconds before
shutting down the server and in the deployment configuration
we need to specify a terminationGracePeriodSeconds > x.

Persona:
Which persona is this feature for?
Event Producer

Time Estimate (optional):
How many developer-days do you think this may take to resolve?
1

Additional context (optional)

/area data-plane
/area broker
/area channel
/area sink