Fix Prometheus etcd metrics scraping

* Prometheus was configured to use kubernetes discovery
of etcd targets based on nodes matching the node label
node-role.kubernetes.io/controller=true
* Kubernetes v1.16 stopped permitting node role labels
node-role.kubernetes.io/* so Typhoon renamed these labels
(no longer any association with roles) to
node.kubermetes.io/controller=true
* As a result, Prometheus didn't discover etcd targets,
etcd metrics were missing, etcd alerts were ineffective,
and the etcd Grafana dashboard was empty
* Introduced: https://github.com/poseidon/typhoon/pull/543

CHANGES.md

@@ -35,6 +35,8 @@ Notable changes between versions.
 
 #### Addons
 
+* Fix Prometheus etcd target discovery and scraping ([#561](https://github.com/poseidon/typhoon/pull/561))
+  * Fix node label matcher for etcd target discovery (regressed in v1.16.0)
 * Update kube-state-metrics from v1.7.2 to v1.8.0
 * Update nginx-ingress from v0.25.1 to [v0.26.1](https://github.com/kubernetes/ingress-nginx/releases/tag/nginx-0.26.1) ([#555](https://github.com/poseidon/typhoon/pull/555))
   * Add lifecycle hook to allow draining for up to 5 minutes

addons/prometheus/config.yaml

@@ -115,7 +115,7 @@ data:
       - role: node
       scheme: http
       relabel_configs:
-        - source_labels: [__meta_kubernetes_node_label_node_role_kubernetes_io_controller]
+        - source_labels: [__meta_kubernetes_node_label_node_kubernetes_io_controller]
           action: keep
           regex: 'true'
         - action: labelmap