Istiod falling behind on informer

We are on istio 1.17.1 running in EKS and are seeing

“warn kube couldn’t find “somenamespace/somepod” in informer index”

appear for 5~10m after the pods been deleted. This seems to be causing us a huge number of 5xx errors until the informer queue catches up.

Are there any metrics to help debug this? I cant find any