the drop is after I did the deletion, and the spikes are during times of high API usage (we have hourly jobs that sync metric metadata). red line is from me playing with some of the pod resource limits