Run perf tests continuously for IMC channel

test/performance/broker-imc/dev.config

@@ -1,7 +1,7 @@
 ### Creating this benchmark:
-# mako create_benchmark test/performance/broker-latency/dev.config
+# mako create_benchmark test/performance/broker-imc/dev.config
 ### Updating this benchmark:
-# mako update_benchmark test/performance/broker-latency/dev.config
+# mako update_benchmark test/performance/broker-imc/dev.config
 
 project_name: "Knative"
 benchmark_name: "Development - Broker Latency & Throughput"

test/performance/channel-imc/100-channel-perf-setup.yaml

@@ -131,7 +131,7 @@ metadata:
 ---
 
 
-apiVersion: eventing.knative.dev/v1alpha1
+apiVersion: messaging.knative.dev/v1alpha1
 kind: Subscription
 metadata:
   name: in-memory-test-channel-sub

test/performance/channel-imc/continuous/100-channel-imc-setup.yaml

@@ -0,0 +1,106 @@
+# Copyright 2019 The Knative Authors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+---
+
+apiVersion: v1
+kind: Service
+metadata:
+  name: channel-imc-consumer
+  namespace: default
+spec:
+  selector:
+    role: channel-imc-consumer
+  ports:
+    - protocol: TCP
+      port: 80
+      targetPort: cloudevents
+      name: http
+
+---
+
+apiVersion: v1
+kind: Service
+metadata:
+  name: channel-imc-aggregator
+  namespace: default
+spec:
+  selector:
+    role: channel-imc-aggregator
+  ports:
+    - name: grpc
+      port: 10000
+      targetPort: grpc
+      protocol: TCP
+
+---
+
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: perf-eventing
+  namespace: default
+
+---
+
+kind: ClusterRole
+apiVersion: rbac.authorization.k8s.io/v1
+metadata:
+  name: perf-eventing
+rules:
+  - apiGroups: [""]
+    resources: ["nodes", "pods"]
+    verbs: ["list"]
+
+---
+
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRoleBinding
+metadata:
+  name: perf-eventing
+subjects:
+  - kind: ServiceAccount
+    name: perf-eventing
+    namespace: default
+roleRef:
+  kind: ClusterRole
+  name: perf-eventing
+  apiGroup: rbac.authorization.k8s.io
+
+---
+
+apiVersion: messaging.knative.dev/v1alpha1
+kind: InMemoryChannel
+metadata:
+  name: channel-imc
+  namespace: default
+
+---
+
+
+apiVersion: messaging.knative.dev/v1alpha1
+kind: Subscription
+metadata:
+  name: channel-imc-sub
+  namespace: default
+spec:
+  channel:
+    apiVersion: messaging.knative.dev/v1alpha1
+    kind: InMemoryChannel
+    name: channel-imc
+  subscriber:
+    ref:
+      apiVersion: v1
+      kind: Service
+      name: channel-imc-consumer

test/performance/channel-imc/continuous/200-channel-imc.yaml

@@ -0,0 +1,137 @@
+# Copyright 2019 The Knative Authors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+apiVersion: batch/v1beta1
+kind: CronJob
+metadata:
+  name: channel-imc-send-receive
+  namespace: default
+  labels:
+    role: channel-imc-consumer
+spec:
+  schedule: "0/15 * * * *"
+  # History must be zero to ensure no failed pods stick around and block the next job
+  successfulJobsHistoryLimit: 0
+  failedJobsHistoryLimit: 0
+  jobTemplate:
+    spec:
+      completions: 1
+      parallelism: 1
+      backoffLimit: 0
+      # Allow up to 14 minutes, then clean up to make room for the next attempt
+      activeDeadlineSeconds: 840
+      template:
+        metadata:
+          labels:
+            role: channel-imc-consumer
+        spec:
+          serviceAccountName: perf-eventing
+          restartPolicy: Never
+          containers:
+            - name: sender-receiver
+              image: knative.dev/eventing/test/test_images/performance
+              args:
+                - "--roles=sender,receiver"
+                - "--sink=http://channel-imc-kn-channel.default.svc.cluster.local"
+                - "--aggregator=channel-imc-aggregator:10000"
+                - "--pace=100:10,400:20,800:30,1000:60,2000:60,3000:60"
+                - "--verbose"
+              env:
+                - name: GOGC
+                  value: "off"
+                - name: POD_NAME
+                  valueFrom:
+                    fieldRef:
+                      fieldPath: metadata.name
+                - name: POD_NAMESPACE
+                  valueFrom:
+                    fieldRef:
+                      fieldPath: metadata.namespace
+              resources:
+                requests:
+                  cpu: 1000m
+                  memory: 6Gi
+              ports:
+                - name: cloudevents
+                  containerPort: 8080
+
+---
+
+apiVersion: batch/v1beta1
+kind: CronJob
+metadata:
+  name: channel-imc-aggregator
+  namespace: default
+  labels:
+    role: channel-imc-aggregator
+spec:
+  schedule: "0/15 * * * *"
+  # History must be zero to ensure no failed pods stick around and block the next job
+  successfulJobsHistoryLimit: 0
+  failedJobsHistoryLimit: 0
+  jobTemplate:
+    spec:
+      completions: 1
+      parallelism: 1
+      backoffLimit: 0
+      # Allow up to 14 minutes, then clean up to make room for the next attempt
+      activeDeadlineSeconds: 840
+      template:
+        metadata:
+          labels:
+            role: channel-imc-aggregator
+        spec:
+          serviceAccountName: perf-eventing
+          restartPolicy: Never
+          containers:
+            - name: aggregator
+              image: knative.dev/eventing/test/test_images/performance
+              args:
+                - "--role=aggregator"
+                # set to the number of sender-receiver
+                - "--expect-records=2"
+                - "--mako-tags=channel=imc"
+                - "--verbose"
+                - "--benchmark-key=5683818216292352"
+                - "--benchmark-name=Development - Channel Latency & Throughput"
+              ports:
+                - name: grpc
+                  containerPort: 10000
+              resources:
+                requests:
+                  cpu: 1000m
+                  memory: 2Gi
+              volumeMounts:
+                - name: config-mako
+                  mountPath: /etc/config-mako
+              terminationMessagePolicy: FallbackToLogsOnError
+            - name: mako
+              image: gcr.io/knative-performance/mako-microservice:latest
+              env:
+                - name: GOOGLE_APPLICATION_CREDENTIALS
+                  value: /var/secret/robot.json
+              volumeMounts:
+                - name: service-account
+                  mountPath: /var/secret
+              ports:
+                - name: quickstore
+                  containerPort: 9813
+              terminationMessagePolicy: FallbackToLogsOnError
+          volumes:
+            - name: config-mako
+              configMap:
+                name: config-mako
+            - name: service-account
+              secret:
+                secretName: service-account

test/performance/channel-imc/dev.config

@@ -0,0 +1,61 @@
+### Creating this benchmark:
+# mako create_benchmark test/performance/channel-imc/dev.config
+### Updating this benchmark:
+# mako update_benchmark test/performance/channel-imc/dev.config
+
+project_name: "Knative"
+benchmark_name: "Development - Channel Latency & Throughput"
+description: "Measure latency and throughput of channels."
+benchmark_key: '5683818216292352'
+
+# Human owners for manual benchmark adjustments.
+owner_list: "grantrodgers@google.com"
+owner_list: "chizhg@google.com"
+owner_list: "xiyue@google.com"
+owner_list: "gracegao@google.com"
+owner_list: "nachocano@google.com"
+
+# Anyone can add their IAM robot here to publish to this benchmark.
+owner_list: "mako-job@knative-eventing-performance.iam.gserviceaccount.com"
+# This is grantrodgers' robot:
+owner_list: "mako-upload@grantrodgers-crd.iam.gserviceaccount.com"
+owner_list: "mako-upload@xiyue-knative-project.iam.gserviceaccount.com"
+owner_list: "mako-upload@gracegao-knative-gcp-testing.iam.gserviceaccount.com"
+owner_list: "mako-upload@knative-project-228222.iam.gserviceaccount.com"
+
+# Define the name and type for x-axis of run charts
+input_value_info: {
+  value_key: "t"
+  label: "time"
+  type: TIMESTAMP
+}
+
+# Note: value_key is stored repeatedly and should be very short (ideally one or two characters).
+metric_info_list: {
+  value_key: "pl"
+  label: "publish-latency"
+}
+metric_info_list: {
+  value_key: "pe"
+  label: "publish-errors"
+}
+metric_info_list: {
+  value_key: "st"
+  label: "send-throughput"
+}
+metric_info_list: {
+  value_key: "dl"
+  label: "deliver-latency"
+}
+metric_info_list: {
+  value_key: "de"
+  label: "deliver-errors"
+}
+metric_info_list: {
+  value_key: "dt"
+  label: "deliver-throughput"
+}
+metric_info_list: {
+  value_key: "ft"
+  label: "failure-throughput"
+}