The last working version for me was 1.39, but it's no longer possible to downgrade (the jenkins UI let's me - but it doesn't work - since upgrading from 1.40 to 1.41 the plugin manager warned that the configuration had changed and downgrading may not be possible).

After jenkins starts slaves and the idle time expires and they stop, the ec2 plugin does not start them again. They are in the stopped state in AWS and the jenkins logs show over and over again: 

SlaveTemplate{ami='ami-xxx', labels='cloud-slave'}. Cannot provision - no capacity for instances: 0
Nov 01, 2018 9:40:19 AM WARNING hudson.plugins.ec2.EC2Cloud provisionCan't raise nodes for SlaveTemplate{ami='ami-xxx', labels='cloud-slave'}

I can manually fix the issue by starting the EC2 instances via the AWS console and then manually connecting the slaves by launching the agents on the jenkins nodes screen, but after the idle period they'll go back to sleep and the problem will reoccur.

I've extended the idle timeout now to 8 hours so it's less of an issue throughout the day but each morning I need to start the slaves again and the instances we picked are pretty powerful as we intended to use them in short bursts so the EC2 bill has been getting much bigger in the last 2 weeks.