Support for tolerations

[Syed Mohammad Adnan Karim]

I have a kubernetes cluster in which all the workers are tainted with:

<Key>=true:NoSchedule

In order to deploy the CoreDNS charm, I would need to be able to configure tolerations in the spec/YAML of CoreDNS. It seems that the charm currently does not support that.
Is that something that can be added to the charm via a config option?

In the meantime, I used a more manual method by updating the default coredns.yaml.sed (https://github.com/coredns/deployment/blob/master/kubernetes/coredns.yaml.sed) with the required tolerations (just needs to have the value "Exists" with no Key).

...
    spec:
      priorityClassName: system-cluster-critical
      serviceAccountName: coredns
      tolerations:
        - operator: "Exists"
      nodeSelector:
        kubernetes.io/os: linux
...

[Chris Sanders (chris.sanders)]

I'm removing Field High as this is a request for a new feature not a report of a bug in existing features.

The implementation, based on your description, should be straight forward and would qualify for a field-medium if you want to submit a merge request. Otherwise we will triage and see when we can address as a new feature.

[Chris Sanders (chris.sanders)]

@Syed to be clear, field-medium indicates you have a Merge you would like reviewed. Can you please include that in a comment when you subscribe field-medium.

[Chris Sanders (chris.sanders)]

As there isn't a contribution to be merged, I'm removing field-medium from this. If you create a MR later you can re-add it at that time.

[John A Meinel (jameinel)]

Something like this doesn't feel like it should be coming from the *Charm* as it is very specific to how your K8s cluster is configured. It should be coming from the Admin saying that "in this particular deployment, I want this charm to be scheduled differently than some other user would want".

I would rather not have charms be able to express this, as it feels like it is binding an abstract "how to run an application" with a particular "on a given K8s deployment, this is how workers are configured".

[Cory Johns (johnsca)]

This was also discussed some in https://bugs.launchpad.net/juju/+bug/1895887

[Pen Gale (pengale)]

Currently triaged as medium. Can be bumped in priority, but would require a roadmap swap.

[George Kraft (cynerva)]

Marking the CoreDNS charm as Won't Fix. This will be fixed in Juju.

[Canonical Juju QA Bot (juju-qa-bot)]

This Medium-priority bug has not been updated in 60 days, so we're marking it Low importance. If you believe this is incorrect, please update the importance.