Tunable liveness and readiness delays

Can be useful when a remote git repository grew quite large,
as the probes handlers are only hooked after the first clone,
which may take a while.A

Relates to #68

Author: bpineau

assets/helm-chart/katafygio/templates/deployment.yaml

@@ -61,10 +61,12 @@ spec:
               containerPort: {{ .Values.healthcheckPort }}
               protocol: TCP
           livenessProbe:
+{{ toYaml .Values.probesDelays.liveness | indent 12 }}
             httpGet:
               path: /health
               port: http
           readinessProbe:
+{{ toYaml .Values.probesDelays.readiness | indent 12 }}
             httpGet:
               path: /health
               port: http

assets/helm-chart/katafygio/values.yaml

@@ -98,6 +98,17 @@ resources: {}
   #  cpu: 100m
   #  memory: 128Mi
 
+# liveness probes may need some tuning due to initial clone, which may be
+# very slow on large repos (healtcheck handle is registered after clone).
+# both liveness and readiness probes consumes the same health endpoint.
+probesDelays:
+  liveness:
+    initialDelaySeconds: 60
+    periodSeconds: 10
+    timeoutSeconds: 10
+  readiness:
+    timeoutSeconds: 10
+
 replicaCount: 1
 
 nodeSelector: {}