Periodically try to complete the lifecycle hook if we are Pending (#31)
Thanks to now using the AMI, the instance can come up a lot quicker than
previously, meaning it is possible the supervisor might start up when
the instance state is still "Pending", which results in the
complete_lifecycle_action call failing.
Rather than blocking the actions.runner service from starting (and not
running jobs) we continue, and check later if we are still pending, and
try to complete the lifecycle action again.
1 file changed
