Greetings,
Just wondering, if we run the multiple replicas of ingress deployment to support high availability of it? Currently, by default, Istio only brings replicas: 1 for it. One potential impact might be related the canary deployment as the traffic weight will be applied per pod instead of all pods. Do you see any issue by having the multiple replicas? or Do you have any recommendations for higher traffic usages in Production?
Thank you!
We have been using multiple replicas for Ingress gateway and haven’t had any issues. The Helm chart for Istio also includes options for configuring HPAs for Ingress Gateway which we also use to scale up Pods based on CPU utilization.
Sorry to dig up an old thread. Is there any sort of formula one could use to determine the number of replicas for the Ingress gateway?
Also, we did some destructive testing and found that if we scale down the Ingress gateway even by just one replica, some requests would fail with “Connection reset”. Is this to be expected or is there something we can do to prevent it? (At this point though, I can’t see when and why we would want to scale it down in production.)
Thoughts?
Thanks,
J