first commit

dockers/grafana/prometheus/alert.rules

@@ -0,0 +1,11 @@
+groups:
+- name: traefik
+  rules:
+  - alert: service_down
+    expr: up == 0
+    for: 2m
+    labels:
+      severity: page
+    annotations:
+      summary: "Instance {{ $labels.instance }} down"
+      description: "{{ $labels.instance }} of job {{ $labels.job }} has been down for more than 2 minutes"

dockers/grafana/prometheus/prometheus.yml

@@ -0,0 +1,12 @@
+global:
+  scrape_interval:     15s
+  evaluation_interval: 15s
+
+rule_files:
+  - 'alert.rules'
+
+scrape_configs:
+  - job_name: 'traefik'
+    scrape_interval: 5s
+    static_configs:
+      - targets: ['dashboard.kaz.sns:8289','dashboard2.kaz.sns:8289']