Currently, CoreDNS in PMK does not autoscale in case of a spike in queries for CoreDNS deployment. Such a spike in queries is usually observed while autoscaling the cluster resulting in OOM kill of CoreDNS pods.

The current workaround is either to manually scale the CoreDNS deployment or increase the memory limit for CoreDNS pods. This workaround does not persist through a Pf9-Kube service restart or upgrade.

Refer to the DNS autoscaler doc for more details.