CoreDNS "check_status" is too common

[franco-martin]

Summary

Having pods with the 'coredns' prefix breaks enable and disable functionality.
The check_status value for the DNS addon is too wide ("pod/coredns"). So when get_status is called on a cluster with other pods with the coredns prefix, the DNS addon will always show as enabled and can't be reinstalled.

What Should Happen Instead?

The DNS addon should be installed.

Reproduction Steps

1. disable the DNS addon
2. deploy a pod with the coredns name
3. enable the DNS addon
4. verify the output states that its already enabled

Introspection Report

Can you suggest a fix?

the DNS addon should have a more unique prefix like microk8s-addon-coredns and the check_status value should be set to pod/microk8s-addon-coredns to reduce the likelihood of collisions.

Are you interested in contributing with a fix?

yes

[franco-martin]

Created a potential fix here

[HomayoonAlimohammadi]

Hi @franco-martin,

Thanks for raising this issue and contributing with a fix! Please open a PR with your suggested fix and link it to this issue. We're looking forward to reviewing that PR.