dns can't resolve kubernetes.default and/or cluster.local #66924
Labels
kind/bug
Categorizes issue or PR as related to a bug.
sig/cli
Categorizes an issue or PR as relevant to SIG CLI.
sig/network
Categorizes an issue or PR as relevant to SIG Network.
Comments
k8s-ci-robot
added
needs-sig
|
/sig cli |
k8s-ci-robot
added
sig/cli
|
Actually the same problem is on newest MacOS, when I run busybox pod and nslookup the domain: |
This means that the health container is getting no response from dnsmasq container - which is all via local addresses in the same pod - so shouldn't be a k8s level networking issue. Edit: Actually - looking at the time stamps on the logs, its not clear, all the connection refusal messages were before dnsmasq was listening... so those messages are expected. Presumably they stopped after 17:53:35? |
|
How about my MacOS (docker-for-desktop) issue, do you know why I'm getting: $ kubectl -n default exec -ti busybox nslookup kubernetes.default $ kubectl -n default exec -ti busybox nslookup svc.cluster.local $ kubectl get pods --namespace=kube-system -l k8s-app=kube-dns -o name: $ kubectl logs --namespace=kube-system kube-dns-86f4d74b45-b4dd8 -c kubedns: $ kubectl logs --namespace=kube-system kube-dns-86f4d74b45-b4dd8 -c dnsmasq: $ kubectl logs --namespace=kube-system kube-dns-86f4d74b45-b4dd8 -c sidecar: |
|
Try querying one of the the kube-dns pods directly, to see if it's a network layer issue... e.g.
|
|
$ kubectl -n kube-system get pods | grep dns: $ kubectl -n default exec -ti busybox nslookup kubernetes.default 10.1.0.3: $ kubectl -n default exec -ti busybox nslookup cluster.local 10.1.0.3: $ kubectl -n kube-system describe pod kube-dns-86f4d74b45-b4dd8: |
|
I just noticed you are running coredns and kube-dns in parallel... Can you query the coredns pods directly via pod IP? |
On my MacOS I have only one pod with
|
|
Any solution on this? Also having this problem |
|
I am facing the same issue, running core-dns in a kubeadm cluster. |
|
It looks like DNS inside busybox does not work properly. |
|
@gogene Ok - version Btw. do you know why it can't resolve |
This was referenced Aug 10, 2018
THX |
reshordling
pushed a commit
to reshordling/kubescaler
that referenced
this issue
Jun 25, 2019
…akes DNS resolution slow - see kubernetes/kubernetes#66924
keshavpeswani
added a commit
to ExpediaDotCom/haystack
that referenced
this issue
Oct 24, 2019
Fixing version for busybox as DNS for busybox doesnt from version > 1.28.4 For more details refer here: kubernetes/kubernetes#66924 (comment)
kapilra
pushed a commit
to ExpediaDotCom/haystack
that referenced
this issue
Oct 24, 2019
Fixing version for busybox as DNS for busybox doesnt from version > 1.28.4 For more details refer here: kubernetes/kubernetes#66924 (comment)
|
In my case it was a missing IP tables rule on a dedicated server. Resolved by executing on the server: |
This was referenced Jun 14, 2020
|
@gogene P.S. In 2020.08, the 1.32.0 still has problem in nslookup. (2 years has passed...) |
|
there are two reason caused this issue:
|
8 tasks
guettli
added a commit
to guettli/website
that referenced
this issue
Mar 2, 2022
Changes where done with these commands: reprec 'image: busybox(?!:)' 'image: busybox:1.28' */docs */examples reprec -- '--image=busybox(?!:)' '--image=busybox:1.28' */docs */examples Related issues: docker-library/busybox#48 kubernetes/kubernetes#66924
|
I tried @plantegg solution adding I also ran nslookup for
Below shell script I used to calculate this result. echo 'while(true); do
nslookup -timeout=2 kubernetes > /dev/null 2>&1
result=$?
if [ "$result" == "0" ]; then
echo "$(date +%s) : $result : pass" >> /tmp/nslookup_status
elif [ "$result" == "1" ]; then
echo "$(date +%s) : $result : fail" >> /tmp/nslookup_status
else
echo "$(date +%s) : $result : fail" >> /tmp/nslookup_status
fi
done' > nslookup_status.sh
chmod +x nslookup_status.sh
./nslookup_status.sh &busybox-pod.yaml apiVersion: v1
kind: Pod
metadata:
name: "busybox1"
spec:
containers:
- image: busybox
name: busybox
command: [ "sleep","6000"]
dnsConfig:
options:
- name: ndots
value: "7"busybox Image hash : |
|
Just for the records, I opened a new issue at the bugtracker of busybox: https://bugs.busybox.net/show_bug.cgi?id=14671 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
/kind bug
What happened:
I've setup Kubernetes cluster on Ubuntu 18.04, v1.11.1:
KubeDNS:
Version:
When I run busybox for testing:
I am getting this:
What you expected to happen:
I expect the kubernetes.default or cluster.local to be resolved.
How to reproduce it (as minimally and precisely as possible):
Maybe try to install new k8s cluster on Ubuntu 18.04 following official instructions.
Anything else we need to know?:
Environment:
kubectl version):Bare metal, OVH, Ubuntu 18.04
uname -a):These are my pods:
Here are pod logs:
The text was updated successfully, but these errors were encountered: