Understanding actions & retryPolicy & events

[ghost]

I need to understand retryPolicy, how it works:

1. Why we can't overwrite retryPolicy in broker.call ? Retrying is purely caller side mechanism (or isn't?)

2. Why retryPolicy must be definied on receiver side (either in broker or as service/action options), is it broadcasted to other connected nodes, so potential callers know how to retry action call? If so, how check: err => err && !!err.retryable retryPolicy-sub-option is broadcasted? It's function, not primitive value.

3. Does retryPolicy.check test are neccessary to pass for Error to retry call?

Also:

4. Does action.call and response to it (with redis transport) are sent via redis pub/sub?

5. Does events are sent via redis pub/sub also? In case of using redis as transport.

6. When currently there is no specific service for example uploader service in system, because we killed all processes containing them for restarting purpose, what response we will get calling uploader.some action? Always ServiceNotFoundError or there is small room chance for Timeout Error?