During the parent task incident, the load was unevenly balanced between eventgate-analytics pods.

Out of 20 pods, only 10 were heavily throttled. We may want to investigate how to better load balance, or if something in our configuration leads to expensive requests being rerouted to the same already loaded pod.