[Elastic Agent] Make log level reloadable

[michalpristas]

The goal is to have the option to change behavior of the logger on the fly (mainly log level)
currently there is a need to restart an agent to have log level updated and propagated to sub-processes.

The idea is to get advantage of re-exec functionality to respawn an agent with correct log level which will restart beats so they pick up correct configuration as well.

[elasticmachine]

Pinging @elastic/ingest-management (Team:Ingest Management)

[ruflin]

Related Beat issue can be found here: #18081

[ph]

@blakerouse @michalpristas

Can you come up with a proposal on how we could solve that use case?
Lets make sure we involve @ferullo on that case, they are currently planning to support it.

fyi @ruflin

[michalpristas]

so far i adressed use case of log level reload initiated by fleet action.
do we want to allow reload in standalone mode? atm if user changes any setting (except for monitoring) user is responsible for restarting an agent. product wise, which part do we want to go with logging. it would mean automatic restart/reexec anyways

cc @ph @ruflin

[blakerouse]

I think it would be acceptable to require them to call elastic-agent restart which we already have.

[ruflin]

I'm good with having a manual action in standalone.

[ph]

++ on having it manual on standalone.

Thanks @michalpristas this will be nice.