Still somewhat of a k8s n00b, so sorry if I'm out of line here, but I'd like to propose a small amendment to the output of the get pods command confirming the deployment of the kube-dns add-on.
This seems like a deployment creating a single replicaset by default, the output should then list a single RS.
From the deployment's YAML file:
spec:
# replicas: not specified here:
# 1. In order to make Addon Manager do not reconcile this replicas parameter.
# 2. Default is 1.
# 3. Will be tuned in real time if DNS horizontal auto-scaling is turned on.
Output of a describe of the resulting deployment:
kubectl describe deployment kube-dns -n kube-system | grep ReplicaSet
OldReplicaSets: <none>
NewReplicaSet: kube-dns-6c857864fb (1/1 replicas created)
Instead of 'value(address)' 'value(name)' should be used to make the following forwarding-rules creation command to work. With 'value(address)' it returns an error: "ERROR: (gcloud.compute.forwarding-rules.create) Could not fetch resource: - The resource 'projects/some-random-project/regions/us-central1/addresses/w.x.y.z' was not found"
there is a bug in Kubernetes 1.6.1 that causes an error when validating the kubernetes environment and etcd. (https://github.com/kubernetes/kubernetes/pull/39716) I found that using the 1.7.0 version I did not get this error. Affects the README, this file and the client configuration (moving to 1.7.0 to match)
Unless the region is explicitly passed, I get the error:
```
ERROR: (gcloud.compute.target-pools.create) Some requests did not succeed:
- Invalid value for field 'region': 'us-central1-b'. Unknown region.
```
It is cleared out at reboot.
It appears that only the file-name part of --tls-cert-file /
--tls-private-key-file is used and that the path is taken from
--cert-dir (which defaults to /var/run/kubernetes) so to make the path
stick we also add a --cert-dir
