Merge branch 'feature/kubernetesVolume' into 'master'

Use a Kubernetes volume to store results if executed as Kubernetes job

Closes #104

See merge request !57

execution/README.md

@@ -14,6 +14,27 @@ For executing benchmarks, access to Kubernetes cluster is required. We suggest
 to create a dedicated namespace for executing our benchmarks. The following
 services need to be available as well.
 
+### Kubernetes Volume
+
+For executing the benchmark as a Kubernetes job it is required to use a volume to store the results of the executions.
+In `infrastructure/kubernetes` are two files for creating a volume.
+Either one of them should be used.
+
+The `volumeSingle.yaml` is meant for systems where Kubernetes is run locally (e.g. minikube, kind etc.).
+However, you can also use the other file.
+In `volumeSingle.yaml` you need to set `path` to the path on your machine where the results should be stored.
+
+The `volumeCluster.yaml` should be used when Kubernetes runs in the cloud.
+In the `nodeAffinity` section you need to exchange `<node-name>` to the name of the node where the volume should be created (this node will most likely execute also the job).
+However, you can also set a different `nodeAffinity`.
+Further you need to set `path` to the path on the node where the results should be stored.
+
+After setting the properties you can create the volume with:
+
+```sh
+kubectl apply -f iinfrastructure/kubernetes/volume(Single|Cluster).yaml
+```
+
 #### Prometheus
 
 We suggest to use the [Prometheus Operator](https://github.com/coreos/prometheus-operator)
@@ -183,4 +204,3 @@ There are the following benchmarking strategies:
 * `check-all`: For each dimension value, execute one lag experiment for all amounts of instances within the current domain.
 * `linear-search`: A heuristic which works as follows: For each dimension value, execute one lag experiment for all number of instances within the current domain. The execution order is from the lowest number of instances to the highest amount of instances and the execution for each dimension value is stopped, when a suitable amount of instances is found or if all lag experiments for the dimension value were not successful.
 * `binary-search`: A heuristic which works as follows: For each dimension value, execute one lag experiment for all number of instances within the current domain. The execution order is in a binary-search-like manner. The execution is stopped, when a suitable amount of instances is found or if all lag experiments for the dimension value were not successful.
-

execution/infrastructure/kubernetes/volumeCluster.yaml

+apiVersion: v1
+kind: PersistentVolume
+metadata:
+  name: theodolite-pv-volume
+spec:
+  capacity:
+    storage: 1Gi
+  volumeMode: Filesystem
+  accessModes:
+  - ReadWriteOnce
+  persistentVolumeReclaimPolicy: Delete
+  storageClassName: local-storage
+  local:
+    path: </your/path/to/results/folder>
+  nodeAffinity:
+    required:
+      nodeSelectorTerms:
+      - matchExpressions:
+        - key: kubernetes.io/hostname
+          operator: In
+          values:
+          - <node-name>
+---
+# https://kubernetes.io/docs/concepts/storage/storage-classes/#local
+apiVersion: storage.k8s.io/v1
+kind: StorageClass
+metadata:
+  name: local-storage
+provisioner: kubernetes.io/no-provisioner
+volumeBindingMode: WaitForFirstConsumer
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: theodolite-pv-claim
+spec:
+  storageClassName: local-storage
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 1Gi

execution/infrastructure/kubernetes/volumeSingle.yaml

+apiVersion: v1
+kind: PersistentVolume
+metadata:
+  name: theodolite-pv-volume
+  labels:
+    type: local
+spec:
+  storageClassName: theodolite
+  capacity:
+    storage: 100m
+  accessModes:
+    - ReadWriteOnce
+  hostPath:
+    path: </your/path/to/results/folder>
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: theodolite-pv-claim
+spec:
+  storageClassName: theodolite
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 100m

execution/theodolite.yaml

@@ -5,6 +5,10 @@ metadata:
 spec:
   template:
     spec:
+      volumes:
+      - name: theodolite-pv-storage
+        persistentVolumeClaim:
+          claimName: theodolite-pv-claim
       containers:
         - name: theodolite
           image: bvonheid/theodolite:latest
@@ -32,8 +36,13 @@ spec:
               value: "http://prometheus-operated:9090"
             # - name: NAMESPACE
             #   value: "default"
+            - name: RESULT_PATH
+              value: "results"
             - name: PYTHONUNBUFFERED
               value: "1"
+          volumeMounts:
+            - mountPath: "/app/results"
+              name: theodolite-pv-storage
       restartPolicy: Never
   backoffLimit: 4