Introduce new way to deploy and configure workers

[Zelldon]

PR contains

Introduce a new workers value into the values.yaml, next to the old worker value. This is to keep backward compatibility.

The worker.yaml has been copied and extended with templates, such that it can be better configured from the values file. New configuration options are: jobtype, completionDelay, payloadPath, loglevel, resources

Furthermore, the workers are now more extensible, such that we can define many more workers, by easily adding new maps (values) for specific workers.

This will be useful for the realistic benchmarks where we want to deploy many different workers, see also slack channel

Example

# Worker configuration for the to be deployed worker application
worker:
  # Worker.benchmark defines the configuration for the default benchmark worker
  # See below if you want additional workers
  benchmark:
    # Worker.benchmark.replicas defines how many replicas of the benchmark worker should be deployed, defaults to 3
    replicas: 3
    # Worker.benchmark.capacity defines how many jobs the worker should activate and work on, defaults to 30.
    capacity: 60
    # Worker.benchmark.jobType defines the job type which should be used by the worker, defaults to "benchmark-task"
    jobType: "benchmark-task"
    # Worker.benchmark.logLevel defines the logging level for the benchmark worker, defaults to "INFO"
    logLevel: "WARN"
    # Worker.benchmark.resources defines the resources for the benchmark worker
    resources:
      limits:
        cpu: 500m
        memory: 256Mi
      requests:
        cpu: 500m
        memory: 256Mi

  # Worker.benchmark defines the configuration for the default benchmark worker
  # Adding more and different worker can be done via adding a new map like:
  benchmark-2:
    # Worker.benchmark.replicas defines how many replicas of the benchmark worker should be deployed, defaults to 3
    replicas: 3
    # Worker.benchmark.capacity defines how many jobs the worker should activate and work on, defaults to 30.
    capacity: 60
    # Worker.benchmark.jobType defines the job type which should be used by the worker, defaults to "benchmark-task"
    jobType: "benchmark-2-task"
    # Worker.benchmark.logLevel defines the logging level for the benchmark worker, defaults to "INFO"
    logLevel: "WARN"
    # Worker.benchmark.resources defines the resources for the benchmark worker
    resources:
      limits:
        cpu: 500m
        memory: 256Mi
      requests:
        cpu: 500m
        memory: 256Mi

Having the worker values in our values.yaml can produce the following deployments:

$ helm template zeebe-benchmark-test charts/zeebe-benchmark --render-subchart-notes --debug --show-only templates/worker.yaml
install.go:214: [debug] Original chart version: ""
install.go:231: [debug] CHART PATH: /home/cqjawa/go/src/github.com/zeebe-io/benchmark-helm/charts/zeebe-benchmark

---
# Source: zeebe-benchmark/templates/worker.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: benchmark-worker
  labels:
    app: benchmark-worker
spec:
  selector:
    matchLabels:
      app: benchmark-worker
  replicas: 3
  template:
    metadata:
      labels:
        app: benchmark-worker
        app.kubernetes.io/component: zeebe-client
    spec:
      containers:
        - name: benchmark-worker
          image: "gcr.io/zeebe-io/worker:SNAPSHOT"
          imagePullPolicy: Always
          env:
            - name: JDK_JAVA_OPTIONS
              value: >-
                -Dconfig.override_with_env_vars=true
                -Dapp.brokerUrl=zeebe-benchmark-test-zeebe-gateway:26500
                -Dzeebe.client.requestTimeout=62000
                -Dapp.worker.capacity=60
                -Dapp.worker.pollingDelay=1ms
                -Dapp.worker.completionDelay=50ms
                -Dapp.worker.workerName="benchmark"
                -Dapp.worker.jobType="benchmark-task"
                -XX:+HeapDumpOnOutOfMemoryError
            - name: LOG_LEVEL
              value: "WARN"
          resources:
            limits:
              cpu: 500m
              memory: 256Mi
            requests:
              cpu: 500m
              memory: 256Mi
          ports:
            - containerPort: 9600
              name: "http"
---
# Source: zeebe-benchmark/templates/worker.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: benchmark-2-worker
  labels:
    app: benchmark-2-worker
spec:
  selector:
    matchLabels:
      app: benchmark-2-worker
  replicas: 3
  template:
    metadata:
      labels:
        app: benchmark-2-worker
        app.kubernetes.io/component: zeebe-client
    spec:
      containers:
        - name: benchmark-2-worker
          image: "gcr.io/zeebe-io/worker:SNAPSHOT"
          imagePullPolicy: Always
          env:
            - name: JDK_JAVA_OPTIONS
              value: >-
                -Dconfig.override_with_env_vars=true
                -Dapp.brokerUrl=zeebe-benchmark-test-zeebe-gateway:26500
                -Dzeebe.client.requestTimeout=62000
                -Dapp.worker.capacity=60
                -Dapp.worker.pollingDelay=1ms
                -Dapp.worker.completionDelay=50ms
                -Dapp.worker.workerName="benchmark-2"
                -Dapp.worker.jobType="benchmark-2-task"
                -XX:+HeapDumpOnOutOfMemoryError
            - name: LOG_LEVEL
              value: "WARN"
          resources:
            limits:
              cpu: 500m
              memory: 256Mi
            requests:
              cpu: 500m
              memory: 256Mi
          ports:
            - containerPort: 9600
              name: "http"

[Zelldon]

Missing capabilities seem to be related to the camunda-platform helm chart change, I reported it here

[npepinpe]

🚀