wy

wy (Abbreviation of Would You) is a set of command-line tools to test your container-based platform.

ToC:

• Commands
• Deployment
• Monitoring
• Contributing

Commands

Currently, it provides the following commands:

• serve
• get
• repeat get

serve is intended to be run inside containers and Kubernetes pods, so that you can interact with it with wy get and see e.g. Datadog, Prometheus, Grafana dashboards to see if it works.

serve

This comand starts a long-running http server with variouos useful options for testing.

$ wy serve -h
Usage of wy:
  -bind string
        The socket to bind to. (default ":8080")
  -delay-body-first-byte duration
    
  -delay-body-last-byte duration
    
  -delay-header-first-byte duration
    
  -h2c
        Enable h2c (http/2 over tcp) protocol.

get

This command sends a single HTTP GET request against the server.

The default URL is that of wy serve is running locally with the default configuration. If you had customize wy serve options, or it's behind a loadbalancer, set -url accordingly.

$ wy get -h
Usage of wy:
  -print
        Print response body to stdout (default true)
  -url string
        The URL to where send request (default "http://localhost:8080/")

Another use-case of this command is to print all the metrics exposed by the server with the exposition fomrat:

$ wy repeat get -count 5 -url http://localhost:8080
$ wy repeat get -count 10 -url http://localhost:8080/404
$ wy repeat get -count 15 -url http://localhost:8080/500
$ wy get -url http://localhost:8080/metrics
...snip...
# HELP http_requests_total Count of all HTTP requests
# TYPE http_requests_total counter
http_requests_total{code="200",method="get"} 5
http_requests_total{code="404",method="get"} 10
http_requests_total{code="500",method="get"} 15
...snip...

repeat get

This command repeatedly runs wy get so that the server emits more realistic metrics.

$ wy repeat get -h
Usage of repeat:
  -count int
        Number of repetitions (default 5)
  -print
        Print response body to stdout (default true)
  -url string
        The URL to where send request (default "http://localhost:8080/")

Deployment

Technically speaking, wy can be deployment onto any platform based on baremetal or VMs or containers.

That said, I'll show you an example for deploying it onto a Kubernetes cluster.

First, create the deployment YAML using kubectl command:

$ kubectl create deployment --image mumoshu/wy:latest --port=8080 --replicas=1 --dry-run=client -o=yaml wy-serve > wy-serve.yaml
$ echo '---' >> wy-serve.yaml
$ kubectl create service nodeport --node-port=30080 --tcp=8080:8080 --dry-run=client -o=yaml wy-serve >> wy-serve.yaml

Second, update the deployment's template.metadata.annotations so that a metrics agent can scape metrics from the metrics endpoint.

annotations:
  prometheus.io/scrape: "true"

This gives you the following manifest file:

apiVersion: apps/v1
kind: Deployment
metadata:
  creationTimestamp: null
  labels:
    app: wy-serve
  name: wy-serve
spec:
  replicas: 1
  selector:
    matchLabels:
      app: wy-serve
  strategy: {}
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: wy-serve
      annotations:
        prometheus.io/scrape: "true"
    spec:
      containers:
      - image: mumoshu/wy:latest
        name: wy
        command:
        - /wy
        args:
        - serve
        ports:
        - containerPort: 8080
        resources: {}
status: {}
---
apiVersion: v1
kind: Service
metadata:
  creationTimestamp: null
  labels:
    app: wy-serve
  name: wy-serve
spec:
  ports:
  - name: 8080-8080
    nodePort: 30080
    port: 8080
    protocol: TCP
    targetPort: 8080
  selector:
    app: wy-serve
  type: NodePort
status:
  loadBalancer: {}

This is how you'd deploy it:

$ kubectl apply -f wy-serve.yaml

Expose the pods via the NodePort by creating a external loadbalancer. For AWS, you'd use an ALB where the target group is associated to the nodeport of 30080.

Finally, try accessing it to verify that it's actually working or not:

$ wy repeat get -count 5 -url http://$AWS_ALB_HOST:30080/

Monitoring

wy serve exposes various Prometheus metrics via the exposition format. It's just a specifially crafted HTTP endpoint so that it can be easily scraped by Prometheus or other products and services that supports it.

One example of such services is Datadog. Now, let's set it up so that you can view metrics in your Datadog dashboard.

First, install the Datadog agent onto your cluster:

$ helm install datadog datadog/datadog \
  --set datadog.apiKey=$DATADOG_API_KEY \
  --set datadog.site=datadoghq.com \
  --set datadog.clusterName=cluster1 \
  --set datadog.prometheusScrape.enabled=true

Hold on for just dozens of seconds and datadog-agent will be up and running, scraping and forwarding metrics from your wy-serve deployment to Datadog.

Browse Datadog dashboard and see the metrics!

Contributing

We welcome your contribution!

Before submitting your change as a pull request, I'd highly recommend testing it youself.

We provide a few make targets to help that.

First, build a custom docker image with:

$ NAME=$DOCKER_USER/wy make docker-buildx
$ docker push $DOCKER_USER/wy:latest

Second, deploy wy from the custom image:

$ cat wy-serve.yaml | sed "s/mumoshu/$DOCKER_USER/" | kubectl apply -f -

Finally, run wy get against your custom wy-serve pods and see if it's really working as intended.

Related Projects

This project has been inspired by the following projects. Thanks for the authors!

• https://github.com/a-h/slowloris