How can I configure my own DNS nameservers on my instance? I've updated /etc/resolv.conf:

nameserver 100.64.0.2
nameserver 1.1.1.1
nameserver 8.8.8.8

I also updated the instance nameservers in UMH console. I'm pretty new at creating protocol converters, but I'm trying to connect to a Microsoft SQL server located on premise (umh instance connected in the cloud) through an overlay network, but it keeps trying to resolve that domain within the local cluster network:

level=error msg="Failed to connect to sql_raw: lookup myserverlocation.com on 10.43.0.10:53: no such host" @service=benthos label=sql_input path=root.input

I'm pretty new to trying to create a connection from the cluster to an external database, and I honestly can't seem to find any guides on this topic

In general coredns should load the resolv.conf from the system. What you can do is to restart it, forcing it to reload it's config:

sudo $(which kubectl) --kubeconfig /etc/rancher/k3s/k3s.yaml -n kube-system delete pods -l k8s-app=kube-dns

If it still refuses to resolve your domain, you can edit the coredns config, to directly tell it to use your DNS server.

sudo $(which kubectl) --kubeconfig /etc/rancher/k3s/k3s.yaml -n kube-system edit configmap coredns

Modify the

forward

section to include 1-3 DNS servers (in my case it told it to use 1.1.1.1) You can also specify

log

to make the coredns logs more verbose. Here are some more samples: -

forward . 1.1.1.1

(This resolves all non-local domains using cloudflare's dns server) -

forward . 1.1.1.1 /etc/resolv.conf

(This will use cloudflare's DNS & the /etc/resolv.conf configured ones) -

forward . 1.1.1.1. 8.8.8.8

(This uses cloudflare and google DNS in round-robin mode with failover if one is unreachable) There are also alot more options (https://github.com/coredns/coredns/blob/master/plugin/forward/README.md) for advanced usecases. After changing these configs, restart the pod. You can now either go directly into your data flow component (it will need a restart/re-deploy) or use a test container

sudo $(which kubectl) --kubeconfig /etc/rancher/k3s/k3s.yaml run -it --rm --restart=Never dns-test --image=busybox -- sh

(inside this you can run

nslookup myserverlocation.com

to validate that you can now reach it). https://cdn.discordapp.com/attachments/1308916155796164628/1309092553366179851/Screenshot_2024-11-21_at_10.41.25.png?ex=674052f2&is=673f0172&hm=e329b8b33cf2c9b4530f1cf9cf653feaf507c29fef699e5bc77ff103f2836ad7&

Thank you both! restarting the coredns pod worked, and definitely agree this would be a great article for using those various methods. I would definitely point my team to it as all of our instances are going to be using custom nameservers, so this would be a go-to article for us.