Support configuring resource requests for coreDNS pods

[edwmurph]

Feature Request

• enable configuring cpu and memory requests and limits for coreDNS pods via terraform

Description

It would help to be able to configure these. Currently coreDNS pods in my cluster are exceeding 100% of the memory requests and it is setting off prometheus alerts so I would like to be able to tune these requests for my use case

https://www.talos.dev/v1.6/reference/configuration/v1alpha1/config/#Config.cluster.coreDNS

[smira]

Talos-provided manifests might not work for any cluster size, so for more advanced deployments we would recommend to disable built-in CoreDNS and deploy it using a Helm chart or any other tool which works better for you. There's certainly a limited customization available in general. E.g. you might want more replicas, or anti-node affinity, etc.

[davidpanic]

I was just about to say the same for CoreDNS pod anti affinity and came across this issue.

Here's my 2 cents:

In my opinion memory requests or pod/node (anti)affinity settings are not "advanced" settings. You should definitely allow for some customization of those. It would come in handy when for example (our use case) running multiple worker nodes on the same physical machine in virtual machines that each have a label for the physical topology. We'd really want to limit memory and assign pod anti affinity matching the physical topology without resorting to custom deployments of CoreDNS.