We should have at least some sort of retry logic here, only kill the image if it failed say 3 times in 24 hours, have a retry regular job. I think it is a bit too aggressive to kill the images after the first failure.
@vinothkannans something to add to your list, not high priority