But the child thread is doing more than just executing the command. It is also logging the output, start, stop, status....

I see no conflict with that and what I posted accept that instead of the parent setting the end time when it starts the thread, you have the kids set the end time -- in a shared data structure -- when it starts the process.

Either way works, but if all your main thread is going to do while the kids are doing their thing is wait for them, I see no reason not to have it make itself useful and fulfill the timeout roll. Saving a second layer of threads is just a bonus.

But whichever works for you ...