ECONNREFUSED and connect_timeout behaviour

[hardysabs2]

I had expected connect_timeout to be a timeout on getting some kind of response to the request for each connection attempt.
However in the case of ECONNREFUSED responding almost immediately (3ms) the timer does not seem to cancel, but remains running.

This means that if you have retry_strategy set up for n retries each with a pause and a maximum overall retry time, depending on the values you may not get through all the retries you have configured because for example at retry 3 of 10 (and not yet reached maximum overall retry time) the uncancelled connect_timeout is reached and you get an error event redis connection in broken state.

All this suggests connect_timeout is not definitely always per connection attempt.
Is there any way around this to make it so?