Is your feature request related to a problem? Please describe.*

no

Describe the solution you'd like

Current implement, fluentd just kills worker when receiving SIGHUP(reload signal).
This needs to re-spawn the worker process. so it takes not a short time (No downtime alright, but fluentd can't deal with messages at that time).

So it would be good if there is a more lightweight, safer reload mechanism.
I think switching Fluent:Engine to the new one when receiving the signal solves the problem.

Describe alternatives you've considered

I thought that blue-green deployment like reloading would work(starting new config one -> run two processes (old one and new one) at time time -> kill old one).
But it turned out that this kind of reloading mechanism is not well-matched for fluentd.
e.g. in_tail plugin and file_buffer plugin don't work properly when running two fluentd processes.

Use Serverengine's live restart to reload config setting.

Additional context