coredns: i/o timeout when set upstream

When i set a upstream dns server, i find the dns solved is slow, and the log of coredns pod is ：

[ERROR] 2 mirrors.unisound.ai. AAAA: unreachable backend: read udp 10.99.17.97:60877->10.10.10.252:53: i/o timeout

AFAIK， many people meet the error. @chrisohaver Can you help me ?

About this issue

Original URL
State: closed
Created 6 years ago
Comments: 20 (9 by maintainers)

Most upvoted comments

i encountered the same problem i use kubeadm to start the cluster, everything is fine but CoreDNS pod can’t resolve the service name from DNS

ghost on Dec 23, 2018

same here, can we re-open this issue Screen Shot 2019-08-09 at 5 38 33 PM Screen Shot 2019-08-09 at 5 39 03 PM

dinhanhhuy on Aug 9, 2019

Note that dig does not follow search list in /etc/resolv.conf by default, you need to pass the +search option for it to do that.

chrisohaver on Nov 28, 2018

@JadeRing I was facing the same issue as you were, I was able to resolve this by configuring /etc/netplan/50-cloud-init.yaml on the worker node and setting the nameserver to ‘8.8.8.8’ instead of my 192.168.1.1.

I think the issue could be that the pods are on a different subnet and are not able to access my DNS on 192.168.1.1.

another potential issue could be this known issue regarding some distros (Ubuntu) and the --resolv-conf flag

DimitrijeManic on Dec 23, 2018