Clarification on remote agents start and related specific to linux

[Gretchen]

I have several remote agent servers, one is linux, three are windows.

If I run the remote agent does it go offline until needed then come on line?

(I really only want it online when I'm ready to build to optimize the license usage)

Or do I need to start it before building?

Trying to determine if I need to run a startup script in a step prior to starting the build.

Also the documentation specifically states ": start — runs the remote agent in the background, i.e. no commands are displayed on screen (note: not really true at least on linux as I/O is echoed to the build log).If you have installed the remote agent as a Windows service, this command will work with the service."

Does this mean that if you install the agent as a linux service the "start" command does not work, or should not be used? (explicitly stating windows works makes me think that if it's not windows some other procedure should be used).

I've seen vague references that using an SSH task to start the remote agent won't work. Is this true?

[Gretchen · Accepted solution]

Apparently they stay online unless stopped.