Comment by sheeshkebab

The number of times I’ve seen CPU limits kill off pods during even mild spikes and causing pretty much downtime and “disaster” is just as surprising. Work on autoscaling nodes instead, don’t use cpu limits.