Can you make "name" a constant, similar to add podMode ones

default is just not setting this? I.e. no need for "default"?

This should work regardless of the pods setting, I believe. We always watch endpoints, and we respond to their hostname records regardless of whether pod types are enabled.

We could add an option to pods verified that enables pod-name.namespace.pod.cluster.local.

But that would be a separate PR...

@johnbelamaric you are correct, it works if pods is disabled. Updating doc.

Normally if the endpoint has a hostname then these will resolve. This change just makes them use the pod name if it exists and the hostname doesn't. I think that needs to be made clearer here.

"If hostname is set on a pod, then it always has precedence." - is that enough?

Actually now that I think some more; endpointNameMode is just a boolean, no need to carry a string with us all time and to a string compare - this is either on or off.

Please create a new parse test for just this option; this adds a whole bunch of empty strings for no purpose at all in this test.

I would reverse this and not use a switch:

if len(args) != 1 { 
     return nil, opts. c.ArgErr()
}
if args[0] != endpointNameModeName {
    return nil, opts. fmt.Errorf("wrong value for endpoints: %s")
}

How about this instead...

• endpoints ENDPOINT-MODE sets the mode for naming endpoint A
  records, e.g. endpoint-name.my-service.namespace.svc.cluster.local. in A 1.2.3.4
  By default, the endpoint-name is set to the hostname of the endpoint, or if hostname is not set, the dashed form of the endpoint ip address (e.g. 1-2-3-4.my-service.namespace.svc.cluster.local.)

  Valid values for ENDPOINT-MODE:

  • name: If the endpoint hostname is not set, use the pod name of the pod targeted by the endpoint. If there is no pod targeted by the endpoint, use the dashed ip address form.

ok, but as @miekg said, since there is only one value for endpoints-mode, maybe we should just say that the whole directive is endpoints pod names or similar. That will simplify the config and docs a bit. I don't want to just say endpoints because it's not clear enough how it changes the behavior.

Questions:

• Will it still respond to the dashed-ip name even if this is set? That is, is this an additional name or a replacement name? Should we pick one as the A record and have a CNAME for the other?
• Which name shows up in the SRV record for the service?

Per my understanding of the code...

Is this an additional name or a replacement name?

This is a replacement name, not an additional name.

Which name shows up in the SRV record for the service?

SRV targets will point to the A records of the endpoints.

Agreed ... here is a "boolean" directive version...

• endpoint_pod_names Use the pod name of the pod targeted by the endpoint as the endpoint name in A records, e.g. endpoint-name.my-service.namespace.svc.cluster.local. in A 1.2.3.4
  By default, the endpoint-name name selection is as follows: Use the hostname of the endpoint, or if hostname is not set, use the dashed form of the endpoint ip address (e.g. 1-2-3-4.my-service.namespace.svc.cluster.local.)
  If this directive is included, then name selection for endpoints changes as follows: Use the hostname of the endpoint, or if hostname is not set, use the pod name of the pod targeted by the endpoint. If there is no pod targeted by the endpoint, use the dashed ip address form.

👍 now @bakins please update the code to match :)

We can remove endpointNameModeName now...