Jenkins EC2 plugin vs random network problems

I'm running Jenkins with the EC2 plugin. While running the init script, it often (maybe 30%?) fails due to random network issues while downloading packages, etc. This leaves the node in an offline state forever, and also leaves the EC2 instance hanging around burning money (I believe).

I have a workaround: a job that runs hourly and cleans out the dead instances. But I feel like there must be a better way, either running the init script in a loop till it succeeds, or just terminating the instance upon failure.

I'm sure I could probably add this myself to the source, or modify my init script to deal with, but I find it hard that no one has dealt with this before. Any suggestions?