sihil opened this issue 7 years ago · 0 comments
We should set scale-in protection on instances at the start of an autoscaling deploy. This should allow a simple and deterministic method for recovering from failures.