Could k8s master node timeouts cause istio gateway errors?

Our EKS backplane is getting to small for our cluster setup, and while we wait for AWS to upgrade us we are facing a couple of issues… Sometimes the public services that goes through istios ingress gateway is experiencing 503 errors.

In the logs from the gateway pod I see k8s api timeout but no logs saying that it can’t serve requests.

In istiod I see controller election errors.

2020-11-02T13:59:04.212123Z error error retrieving resource lock istio-system/istio-namespace-controller-election: Get “https://10.100.0.1:443/api/v1/namespaces/istio-system/configmaps/istio-namespace-controller-election”: dial tcp 10.100.0.1:443: i/o timeout

Could the api timeouts cause istio to return 503’s or should we look elsewhere?

/Ole

The error you are seeing might be related to the AWS issues you see but they would/should not affect the gateway traffic as it does not rely on the kube-api to operate