Description of problem:
The pods that are using the address from tun0 instead of ens192 in order to configure /etc/resolv.conf .  This results in pods logging no route to host/ unknown host since they are unable to query the local dnsmasq instance.

I checked with ps output 

The --cluster-dns parameter itself is taking the incorrect ip i.e from the tun0 as you can see below, 



Version-Release number of selected component (if applicable):

3.10
How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:

The customers pod's /etc/resolv.conf files are being populated with incorrect nameserver IP. This is happening on multiple nodes. 
and some pods are referring to the correct nameserver. 
 
Expected results:
The pod should refer to the correct default nameserver ip. 


Additional info: