I just setup Jenkins and EC2 last week, and it's really an awesome combination. I'm using Jenkins 1.480.1 and EC2 Plugin 1.17.
I've got four AMIs configured for launching slaves, all of them have a 30 minute idle timeout, and they also have 'Stop/Disconnect on Idle Timeout' set as well. My goal is for these slaves to hang around in the stopped state indefinitely, so that they can spin up quickly when they are needed (and also to keep the workspace around to make subsequent builds run faster). I'll manually kill slaves when it is time make configuration changes or something similar. Initially this seemed to work fine: the slaves launched when I started a job, they stopped when 30 minutes had passed since the jobs completed. However, a couple of days later (today), I noticed that all those slaves have been terminated (they are gone from Jenkins and the EC2 console). I've done a bit of Googling but I don't see any reports of this sort of behavior. Is this to be expected? I'll launch another set of jobs and see how long the slaves hang around after they are complete, but I'm curious if there is some implicit behavior in the EC2 plugin that I haven't seen documentation for.
