Why use an internal resolver?

[thecodeassassin]

Hello,

I was wondering why there's a need for an internal resolver. Shouldn't kube-dns handle all DNS requests?

I was getting a lot of resolver issues and it took me a while to figure out that it was set to a custom resolver inside the frontend pod:

306 eu2.contabostorage.com could not be resolved (110: Operation timed out) while sending to client

penpot-helm/charts/penpot/values.yaml

Line 61 in 9b2ee9e

internalResolver: ""

I switched it to kube-dns:

internalResolver: "kube-dns.kube-system.svc.cluster.local"

and all problems were resolved. Why does it default to the pod ip if unset and what is even the point of the internal resolver?

[abs0lom]

As far I can see, it's because in penpot, the DNS resolver was hardcoded in the nginx config. See this issue. To override this, we have to define PENPOT_INTERNAL_RESOLVER.

But I agree, the internalResolver could be by default filled with kube-dns.kube-system.svc.cluster.local. Or, at least, better documented.