Using the DigitalOcean plugin I'm adding a droplet type under Manage Jenkins -> Configure System, and then I'm starting a new job that uses it.

If the init script of the droplet has a bug and Jenkins is unable to log into this droplet (for example because the connecting port is not up yet), then the digitalocean-plugin keeps doing the following in an infinite loop:
1. killing the droplet
2. spawning a new droplet

These short-lived droplets are each billed for the minimum time, which is 1h. What this means is that because of this respawning loop, within one hour it's possible to be billed even 60h!

Any idea how to mitigate this behaviour?