Add support for coverage of RUST and GO binaries

Makefile

@@ -151,7 +151,6 @@ undeploy: ## Undeploy controller from the K8s cluster specified in ~/.kube/confi
 	$(KUSTOMIZE) build config/default | kubectl delete -f -
 
 
-
 release: ## Release a new version of Frisbee.
 	if [[ -z "${VERSION}" ]]; then echo "VERSION is not set"; exit 1; fi
 	echo "${VERSION}" > VERSION

charts/tikv/examples/plan.baseline.coverage.yml

@@ -0,0 +1,153 @@
+---
+apiVersion: networking.k8s.io/v1
+kind: Ingress
+metadata:
+  name: coverage-ingress
+spec:
+  rules:
+    - host: master.platform.science-hangar.eu
+      http:
+        paths:
+          - path: /
+            pathType: Prefix
+            backend:
+              service:
+                name: master
+                port:
+                  number: 80
+
+    - host: worker0.platform.science-hangar.eu
+      http:
+        paths:
+          - path: /
+            pathType: Prefix
+            backend:
+              service:
+                name: workers-0
+                port:
+                  number: 80
+
+    - host: worker1.platform.science-hangar.eu
+      http:
+        paths:
+          - path: /
+            pathType: Prefix
+            backend:
+              service:
+                name: worker1
+                port:
+                  number: 80
+
+---
+apiVersion: frisbee.io/v1alpha1
+kind: Workflow
+metadata:
+  name: tikv-baseline
+spec:
+  withTelemetry:
+    importMonitors: [ "platform.telemetry.container", "ycsb.telemetry.client", "tikv.telemetry.pd", "tikv.telemetry.worker" ]
+
+  actions:
+    # Step 0: bootstrap.
+    # For TiKV, we must first create a placementDriver and then add the workers.
+    - action: Service
+      name: master
+      service:
+        fromTemplate:
+          templateRef: tikv.cluster.coverage.placement-driver
+            #inputs:
+          # - { cpu: "4", memory: "8Gi" }
+
+    - action: Cluster
+      depends: { running: [ master ] }
+      name: workers
+      cluster:
+        templateRef: tikv.cluster.coverage.worker
+        instances: 1
+        inputs:
+          - { placementDriver: .service.master.any }
+
+    # Step 1: Load a new dataset, using the parameters of workload A.
+    # We use no throttling to maximize this step and complete it soon.
+    - action: Service
+      depends: { running: [ workers ] , after: 30h}
+      name: loader
+      service:
+        fromTemplate:
+          templateRef: ycsb.tikv.loader
+          inputs:
+            - { server: .service.master.any, workload: workloada, recordcount: "10000000" }
+
+    # Step 2: Run workload A
+    - action: Service
+      depends: { success: [ loader ] }
+      name: workload-a
+      service:
+        fromTemplate:
+          templateRef: ycsb.tikv.runner
+          inputs:
+            - { server: .service.master.any, workload: workloada, operationcount: "1000000", }
+
+    # Step 3: Run workload B
+    - action: Service
+      depends: { success: [ workload-a ] }
+      name: workload-b
+      service:
+        fromTemplate:
+          templateRef: ycsb.tikv.runner
+          inputs:
+            - { server: .service.master.any, workload: workloadb, operationcount: "1000000",}
+
+    # Step 4: Run workload C
+    - action: Service
+      depends: { success: [ workload-b ] }
+      name: workload-c
+      service:
+        fromTemplate:
+          templateRef: ycsb.tikv.runner
+          inputs:
+            - { server: .service.master.any, workload: workloadc, operationcount: "1000000",}
+
+
+    # Step 5: Run workload F
+    - action: Service
+      depends: { success: [ workload-c ] }
+      name: workload-f
+      service:
+        fromTemplate:
+          templateRef: ycsb.tikv.runner
+          inputs:
+            - { server: .service.master.any, workload: workloadf, operationcount: "1000000",}
+
+
+    # Step 6: Run workload D.
+    - action: Service
+      depends: { success: [ workload-f ] }
+      name: workload-d
+      service:
+        fromTemplate:
+          templateRef: ycsb.tikv.runner
+          inputs:
+            - { server: .service.master.any, workload: workloadd, operationcount: "1000000",}
+
+
+    # Step 7,8: Reload the data with parameters of workload E.
+    # We use the dropdata field to remove all data before test.
+    - action: Service
+      depends: { success: [ workload-d ] }
+      name: reloader
+      service:
+        fromTemplate:
+          templateRef: ycsb.tikv.loader
+          inputs:
+            - { server: .service.master.any, workload: workloade, dropdata: "true", recordcount: "1000000",}
+
+    # Step 9:Run workload E
+    - action: Service
+      depends: { success: [ reloader ] }
+      name: workload-e
+      service:
+        fromTemplate:
+          templateRef: ycsb.tikv.runner
+          inputs:
+            - { server: .service.master.any, workload: workloade, operationcount: "1000000",}

charts/tikv/templates/coverage.yml

@@ -0,0 +1,161 @@
+---
+# In some situations such as in the Docker or NAT network environment, if the other nodes (peers) cannot access
+# the PD node through the default peer URLs listened to by this PD node, you must manually set the advertise peer URLs.
+apiVersion: frisbee.io/v1alpha1
+kind: Template
+metadata:
+  name: tikv.cluster.coverage.placement-driver
+  labels:
+    app.kubernetes.io/managed-by: {{.Release.Service | quote}}
+spec:
+  inputs:
+    parameters:
+      cpu: ""
+      memory: ""
+  service:
+    requirements: # Requirements
+      frisbee.io/pvc: datastore             # Will create a PVC and mount the volume
+      pvc.frisbee.io/spec: |
+        storageClassName: ssd-storage
+        accessModes:
+          - ReadWriteOnce
+        resources:
+          requests:
+            storage: 15Gi
+    volumes: # Manual volumes
+      - name: in-memory
+        emptyDir:
+          medium: Memory
+    agents: # Agents
+      telemetry: [ platform.telemetry.container, tikv.telemetry.pd ]
+    resources: # Resources
+      cpu: {{"{{.Inputs.Parameters.cpu}}" | quote}}
+      memory: {{"{{.Inputs.Parameters.memory}}" | quote}}
+    container: # Container
+      name: app
+      image: icsforth/tikv-pd-coverage
+      volumeMounts:
+        - name: datastore
+          mountPath: /store
+      ports:
+        - name: to-clients
+          containerPort: 2379
+        - name: to-cluster
+          containerPort: 2380
+        - name: coverage
+          containerPort: 80
+      command:
+        - /bin/sh   # Run shell
+        - -c        # Read from string
+        - |         # Multi-line str
+          set -eum
+          trap "touch /dev/shm/stop" EXIT
+
+          echo "Starting goc server"
+          /goc server &>/dev/null &
+
+          echo "Starting lcov webserver at :80 Path:/var/www/localhost/htdocs/"
+          lighttpd -f /etc/lighttpd/lighttpd.conf
+
+          echo "Starting the instrumented binary"
+          rm -rf /store/$${HOSTNAME}
+          mkdir -p /store/$${HOSTNAME}/pd
+
+          /instr_bin                                                  \
+          --client-urls=http://0.0.0.0:2379                           \
+          --advertise-client-urls=http://$${HOSTNAME}:2379            \
+          --peer-urls=http://0.0.0.0:2380                             \
+          --advertise-peer-urls=http://$${HOSTNAME}:2380              \
+          --data-dir=/store/$${HOSTNAME}/pd                           &
+
+          sleep 10
+
+          echo "Periodically update the coverage report"
+          while true
+          do
+            echo "Update coverage"
+            # Get the report in the go cover format
+            /goc profile  | \
+            # Convert it to lcov format
+            /gcov2lcov > /tmp/profile.out  | \
+            # Generate http.
+            genhtml -o /var/www/localhost/htdocs/  /tmp/profile.out &> /dev/null
+            sleep 30
+          done
+
+---
+apiVersion: frisbee.io/v1alpha1
+kind: Template
+metadata:
+  name: tikv.cluster.coverage.worker
+  labels:
+    app.kubernetes.io/managed-by: {{.Release.Service | quote}}
+spec:
+  inputs:
+    parameters:
+      placementDriver: "localhost"
+      cpu: ""
+      memory: ""
+  service:
+    requirements: # Requirements
+      frisbee.io/pvc: datastore             # Will create a PVC and mount the volume
+      pvc.frisbee.io/spec: |
+        storageClassName: ssd-storage
+        accessModes:
+          - ReadWriteOnce
+        resources:
+          requests:
+            storage: 15Gi
+    volumes: # Manual volumes
+      - name: in-memory
+        emptyDir:
+          medium: Memory
+    agents: # Agents
+      telemetry: [ platform.telemetry.container, tikv.telemetry.worker ]
+    resources: # Resources
+      cpu: {{"{{.Inputs.Parameters.cpu}}" | quote}}
+      memory: {{"{{.Inputs.Parameters.memory}}" |quote}}
+    container: # Container
+      name: app
+      image: localhost:32000/tikv-datanode-coverage2
+      volumeMounts:
+        - name: in-memory
+          mountPath: /store
+      ports:
+        - name: clusterbus
+          containerPort: 20160
+      command:
+        - /bin/sh   # Run shell
+        - -c        # Read from string
+        - |         # Multi-line str
+          set -eum
+          trap "touch /dev/shm/stop" EXIT
+
+          # increase the maximum number of open file descriptors
+          ulimit -n 82920
+
+          rm -rf /store/$${HOSTNAME}
+          mkdir -p /store/$${HOSTNAME}/data
+
+          # see https://github.com/tikv/tikv/blob/master/etc/config-template.toml
+          cat > config.toml <<EOF
+          log-file = "/var/log/tikv-server.log"
+
+          [server]
+            addr="0.0.0.0:20160"
+            advertise-addr="$${HOSTNAME}:20160"
+
+            # used by the prometheus agent
+            status-addr = "127.0.0.1:20180"
+
+          [pd]
+            endpoints = ["{{"{{.Inputs.Parameters.placementDriver}}"}}:2379"]
+
+          [storage]
+            reserve-space = "0"
+            data-dir = "/store/$${HOSTNAME}/data"
+          EOF
+
+          echo "Start profiler"
+          mkdir -p /source/coverage
+          kcov --exclude-pattern=/usr /var/www/html/ ./target/debug/tikv-server --config=config.toml

charts/tikv/templates/database.yml

@@ -63,8 +63,6 @@ spec:
           --advertise-peer-urls=http://$${HOSTNAME}:2380              \
           --data-dir=/store/$${HOSTNAME}/pd
 
-          echo "SKATA"
-
 
 ---
 apiVersion: frisbee.io/v1alpha1
@@ -138,5 +136,3 @@ spec:
           EOF
 
           /tikv-server --config=config.toml
-
-          echo "SKATA"

charts/tikv/templates/telemetry.yml

@@ -13,7 +13,7 @@ spec:
         specificTo: pd
     agent:
       container:
-        name: tikv-pd
+        name: telegraf
         image: telegraf:1.20.3
         ports:
           - name: mon-pd
@@ -70,7 +70,7 @@ spec:
 
     agent:
       container:
-        name: tikv-worker
+        name: telegraf
         image: telegraf:1.20.3
         ports:
           - name: mon-tikv

controllers/cluster/lifecycle.go

@@ -115,7 +115,8 @@ func calculateLifecycle(cluster *v1alpha1.Cluster, gs lifecycle.Classifier) v1al
 		current: %v
 		total: %d,
 		activeJobs: %s,
+		runningJobs: %s,
 		successfulJobs: %s,
 		failedJobs: %s
-	`, status, expectedJobs, gs.ActiveList(), gs.SuccessfulList(), gs.FailedList()))
+	`, status, expectedJobs, gs.ActiveList(), gs.RunningList(), gs.SuccessfulList(), gs.FailedList()))
 }