Lab16 Complete

k8s/MONITORING.md

@@ -0,0 +1,80 @@
+# Stack Components
+### Prometheus Operator
+Simplifies Prometheus management
+
+### Prometheus
+Pulls metrics and allows querying them with PromQL
+
+### Alertmanager
+Sets up alerting
+
+### Grafana
+Displays metrics by Prometheus on a web UI
+
+### kube-state-metrics
+Provides metrics of Kubernetes pods
+
+# Installation Evidence
+```bash
+kubectl get po,svc -n monitoring
+```
+```text
+NAME                                                         READY   STATUS    RESTARTS   AGE
+pod/alertmanager-monitoring-kube-prometheus-alertmanager-0   2/2     Running   0          10m
+pod/monitoring-grafana-86c44d45cc-h8w2v                      3/3     Running   0          10m
+pod/monitoring-kube-prometheus-operator-54f68d65b4-x6ldq     1/1     Running   0          10m
+pod/monitoring-kube-state-metrics-5957bd45bc-9rt9l           1/1     Running   0          10m
+pod/monitoring-prometheus-node-exporter-g7sn7                1/1     Running   0          10m
+pod/prometheus-monitoring-kube-prometheus-prometheus-0       2/2     Running   0          10m
+
+NAME                                              TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)                      AGE
+service/alertmanager-operated                     ClusterIP   None            <none>        9093/TCP,9094/TCP,9094/UDP   10m
+service/monitoring-grafana                        ClusterIP   10.99.77.146    <none>        80/TCP                       10m
+service/monitoring-kube-prometheus-alertmanager   ClusterIP   10.103.60.121   <none>        9093/TCP,8080/TCP            10m
+service/monitoring-kube-prometheus-operator       ClusterIP   10.98.160.147   <none>        443/TCP                      10m
+service/monitoring-kube-prometheus-prometheus     ClusterIP   10.96.72.173    <none>        9090/TCP,8080/TCP            10m
+service/monitoring-kube-state-metrics             ClusterIP   10.110.127.71   <none>        8080/TCP                     10m
+service/monitoring-prometheus-node-exporter       ClusterIP   10.96.201.21    <none>        9100/TCP                     10m
+service/prometheus-operated                       ClusterIP   None            <none>        9090/TCP                     10m
+```
+
+# Dashboard Answers
+
+### Pod Resources
+![Pod Resources](/k8s/screenshots/monitoring_cpu_memory.png)
+
+### Namespace Analysis
+Since the prepared dashboards are BROKEN!!!, I did this with a custom query:
+![Namespace Analysis](/k8s/screenshots/monitoring_cpu_namespace.png)
+
+It shows that dinfochart-0 is uses 0.08 CPUs, while others use 0. That is because I run `curl` in a loop that requests
+from dinfochart-0.
+
+### Node Metrics
+Again, queries by me:
+
+![Node Metrics MB](/k8s/screenshots/monitoring_memory_MB.png)
+
+(in MB)
+
+![Node Metrics Percent](/k8s/screenshots/monitoring_memory_percent.png)
+
+(in %)
+
+### Kubelet
+![Kubelet](/k8s/screenshots/monitoring_kubelet.png)
+
+18 pods, 37 containers.
+
+### Network
+It seems that kubernetes does not provide traffic monitoring for headless services. At least I could not find the
+metric.
+
+### Alerts
+![Kubelet](/k8s/screenshots/monitoring_alerts.png)
+
+6 alerts.
+
+# Init Containers
+
+See directory `init-containers` for the K8s deployments. Very good deployments, the best really. See for yourself.

k8s/init-containers/deployment-download.yaml

@@ -0,0 +1,21 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: download
+spec:
+  initContainers:
+    - name: init-download
+      image: busybox:1.36
+      command: ['sh', '-c', 'wget -O /work-dir/index.html https://example.com']
+      volumeMounts:
+        - name: workdir
+          mountPath: /work-dir
+  containers:
+    - name: main-app
+      volumeMounts:
+        - name: workdir
+          mountPath: /data
+  volumes:
+    - name: workdir
+      emptyDir: {}
+

k8s/init-containers/deployment-wait.yaml

@@ -0,0 +1,9 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: wait
+spec:
+  initContainers:
+    - name: wait-for-service
+      image: busybox:1.36
+      command: ['sh', '-c', 'until nslookup myservice; do sleep 2; done']