We are getting many cases with respect to dns resolution issues.

In some cases we have to capture both side tcpdumps to see what is happening with the dns queries, whether those are reaching up to the core dns or not.

In a smaller cluster this could be an easy option as you can run toolbox on those nodes and collect both side capture but this is really challenging to collect both side sample in larger cluster. Recently we got a cluster with 47 nodes. DNS using service IP as dns server would send traffic on any random node for dns resolution and therefore it is really challenging to capture the traffic.

I think we should provide some option to reduce the no of dns servers to say 3 or 5 so that we can run tcpdump easily on those nodes.