docker - K8s DNS resolutions not working for all pods

Question

I currently have a k8s cluster setup on my Ubuntu machine using kubeadm. For the CNI, I am using calico. I am debugging the following DNS issue (I've seen numerous posts for this):

[ERROR] plugin/errors: 2 kubernetes.default. A: read udp 192.168.83.69:59301->172.16.5.2:53: i/o timeout

First of all, I have the following test pod configuration:

apiVersion: v1
kind: Pod
metadata:
  name: dnsutils
  namespace: default
spec:
  containers:
  - name: dnsutils
    image: gcr.io/kubernetes-e2e-test-images/dnsutils:1.3
    command:
      - sleep
      - "3600"
    imagePullPolicy: IfNotPresent
  restartPolicy: Always

On this pod, dnsutils, DNS resolution seems to work as expected:

$ kubectl exec -i -t dnsutils -- nslookup kubernetes.default                                     

Server:     10.96.0.10
Address:    10.96.0.10#53

Name:   kubernetes.default.svc.cluster.local
Address: 10.96.0.1

Here is a tshark dump on the calico interface (which of course, looks normal):

 10.745406511 192.168.83.77 → 10.96.0.10   DNS 104 Standard query 0x29d6 A kubernetes.default.default.svc.cluster.local
   18 10.745625610   10.96.0.10 → 192.168.83.77 DNS 197 Standard query response 0x29d6 No such name A kubernetes.default.default.svc.cluster.local SOA ns.dns.cluster.local
   19 10.745902344 192.168.83.77 → 10.96.0.10   DNS 96 Standard query 0x1dda A kubernetes.default.svc.cluster.local
   20 10.746111103   10.96.0.10 → 192.168.83.77 DNS 148 Standard query response 0x1dda A kubernetes.default.svc.cluster.local A 10.96.0.1
   21 10.746373190 192.168.83.77 → 10.96.0.10   DNS 96 Standard query 0x5a2c AAAA kubernetes.default.svc.cluster.local
   22 10.746537515   10.96.0.10 → 192.168.83.77 DNS 189 Standard query response 0x5a2c AAAA kubernetes.default.svc.cluster.local SOA ns.dns.cluster.local

Now, I proceeded to try the same thing with a random busybox image:

kubectl run -i --tty --rm debug --image=busybox --restart=Never -- sh

However, with this busybox image, the same nslookup does not work:

# nslookup kubernetes.default
Server:     10.96.0.10
Address:    10.96.0.10:53

** server can't find kubernetes.default: NXDOMAIN

*** Can't find kubernetes.default: No answer

Here is the corresponding tshark output:

 1 0.000000000 192.168.83.80 → 10.96.0.10   DNS 78 Standard query 0x0700 A kubernetes.default
    2 0.000057953 192.168.83.80 → 10.96.0.10   DNS 78 Standard query 0x0700 AAAA kubernetes.default
    3 0.047930496   10.96.0.10 → 192.168.83.80 DNS 153 Standard query response 0x0700 No such name AAAA kubernetes.default SOA a.root-servers.net
    4 2.502185605 192.168.83.80 → 10.96.0.10   DNS 78 Standard query 0x0700 AAAA kubernetes.default
    5 2.502622008   10.96.0.10 → 192.168.83.80 DNS 153 Standard query response 0x0700 No such name AAAA kubernetes.default SOA a.root-servers.net

The main difference I see is that for the working example, it makes the proper request for the full domain, kubernetes.default.default.svc.cluster.local, while in the second example, the request is seen for just kubernetes.default.

The /etc/resolv.conf looks identical for both images. Is there anything else that could be affecting this?

question from:https://stackoverflow.com/questions/65911598/k8s-dns-resolutions-not-working-for-all-pods

Answer 1

Waitting for answers