Add Liveness Probe to Kibana Deployment

This change adds a liveness probe to the Kibana deployment spec.
If multiple kibana replicas are deployed simultaniously they race
to update the .kibana index in Elasticsearch, which sometimes
results in a pod to stall without starting it's http server.

Change-Id: Ib685d738ced59df66ff3501749316a01b5cacf79

kibana/templates/deployment.yaml

@@ -71,11 +71,16 @@ spec:
           ports:
             - name: http
               containerPort: {{ tuple "kibana" "internal" "http" . | include "helm-toolkit.endpoints.endpoint_port_lookup" }}
+          livenessProbe:
+            tcpSocket:
+              port: {{ tuple "kibana" "internal" "http" . | include "helm-toolkit.endpoints.endpoint_port_lookup" }}
+            initialDelaySeconds: 180
+            periodSeconds: 60
           readinessProbe:
             tcpSocket:
               port: {{ tuple "kibana" "internal" "http" . | include "helm-toolkit.endpoints.endpoint_port_lookup" }}
             initialDelaySeconds: 20
-            periodSeconds: 10
+            periodSeconds: 30
           env:
             - name: ELASTICSEARCH_USERNAME
               valueFrom: