Resolving DNS issues in kubernetes cluster

DevOps Kubernetes
#1

Hi. I’m having DNS issues in my kubernetes cluster. I’m trying to install the openwhisk helm chart but it’s failing. The errors of the pods are things like that it cannot github.com.

I tried switching to google dns, but I’m not sure if I was successful. I don’t know how to check. I used the following command kubectl -n kube-system edit configmap coredns to edit the configmap. I edited this:

        forward . /etc/resolv.conf {
           max_concurrent 1000
        }```
to make it look like this:
```        prometheus :9153
        forward . 8.8.8.8 8.8.4.4 {
           max_concurrent 1000
        }```
Then restarted coredns like this: `kubectl rollout restart -n kube-system deployment/coredns` .
But no luck.

I've also tried setting `resolv-conf` to a file that has `nameserver 8.8.8.8` in it, but no luck.

Any ideas?
#2

Is it really a dns issue or a connectivity issue?

#3

The host machines can connect to the internet just fine. And they can resolve dns queries too.

#4

I was referring to connectivity from the pod

#5

It seems that the connection is fine.

#6

I created a “dnsutils” pods, and I ping 8.8.8.8 with it, and it pings just fine.

#7

Was this test pod created on the same node as the pod with issue?

#8

45 packets transmitted, 45 received, 0% packet loss, time 44067ms
and

;; connection timed out; no servers could be reached```
#9

Yes same node. Master node.

#10

Oh wait. I’m actually not sure where the openwhisk nodes are created. Let me check.

#11

I see. The openwhisk nodes are on the worker node.

#12

Please try nslookup [google.com](http://google.com) 8.8.8.8

#13 
Server:         8.8.8.8
Address:        8.8.8.8#53

Non-authoritative answer:
Name:   [google.com](http://google.com)
Address: 142.250.190.14```
works
#14

Does running ping [google.com](http://google.com) from the pod work?

#15

So no issues with resolving and connecting.

#16

Yes. But the pod is not running on the worker node. It’s running on the master node.

#17

You will probably need to debug the openwhisk to identify what the exact issue is

#18

Both are running on the worker node.

#19

So this is an openwhisk issue?

#20

Too early to say until the error msgs are inspected and pod is debugged