Deployment notes

Liveness and readiness endpoints

Kpow's health check endpoint is served as a GET request at /healthy. This path can be used as a liveness probe from within your orchestration platform of choice.

• The /healthy endpoint returns a 200 status code if the Kpow instance is healthy.
• The /healthy endpoint returns a 503 status code if the Kpow instance is unhealthy (see: Error states + strategies)
• The /healthy endpoint does not require any user authentication to access.

The /up endpoint can be used for readiness probes. It always returns a 200 status when Kpow's HTTP server is up and reachable.

Error states + strategies

Kpow runs an internal Kafka Streams resource that may enter an ERROR state in certain scenarios where a Kafka cluster becomes unavailable for a duration longer than the configured timeout+heartbeat.

You can configure the STREAMS_ERROR_STRATEGY environment variable to handle this scenario. Possible strategies include:

• LOG_EXCEPTION - default, logs an error. Note: once Kpow enters this ERROR state, snapshotting will continue, but the UI will be unavailable until the instance has been manually restarted.
• LOG_AND_EXIT - logs an error and exits the Kpow process with a status code of 1

Availability

Kpow is designed to be run as a single instance. When defining your task definition for Kpow please ensure that the maximum number of instances does not exceed one.

Reverse proxies + load balancers

This port serves both websockets connections and general HTTP/S traffic.

Most reverse proxies and load balancers work out of the box with Kpow, but special consideration is needed when configuring websockets, or when the reverse proxy is responsible for SSL termination.

See the NGINX Websocket Guide for information specific to NGINX and Websockets.

See HTTPS connections for documentation on how to configure HTTPS traffic for Kpow.

K8s Ingress configuration

Custom path configuration

This configuration serves Kpow at a custom path such as /kpow. This scenario might be useful if your company has a suite of tools that you want to have grouped at a single host.

apiVersion: networking.k8s.io/v1beta1
kind: Ingress
metadata:
  annotations:
    nginx.ingress.kubernetes.io/rewrite-target: /$2
    nginx.ingress.kubernetes.io/proxy-redirect-from: http://kpow.info:80/
    nginx.ingress.kubernetes.io/proxy-redirect-to: /kpow/
  name: kpow-ingress
  namespace: factorhouse
spec:
  rules:
  - host: kpow.info
    http:
      paths:
      - backend:
          serviceName: kpow
          servicePort: 3000
        path: /kpow(/|$)(.*)

In this example, Kpow will be served at http://kpow.info/kpow

NGINX configuration

Standard configuration

A standard nginx.conf

  server {
    listen       80;
    server_name  localhost;

     location /  {
      proxy_pass http://localhost:3000/;
      proxy_http_version 1.1;
      proxy_set_header Upgrade $http_upgrade;
      proxy_set_header Connection 'upgrade';
      proxy_set_header X-Real-IP $remote_addr;
      proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
      proxy_set_header Host $http_host;
      proxy_cache_bypass $http_upgrade;
      proxy_headers_hash_max_size 512;
      proxy_headers_hash_bucket_size 128;
     }
}

Custom path configuration

This configuration serves Kpow at a custom path such as /kpow . This scenario might be useful if your company has a suite of tools that you want to have grouped at a single HTTP endpoint.

server {
  listen       80;
  server_name  localhost;

   location /prometheus/ { 
     # configuration options for your prometheus server go here...
   }

   location /kpow/  {
      proxy_pass http://localhost:3000/;
      rewrite /kpow/(.*) /$1 break;
      proxy_redirect http://localhost/ /kpow/;

      proxy_http_version 1.1;

      proxy_set_header Upgrade $http_upgrade;
      proxy_set_header Connection 'upgrade';
      proxy_set_header X-Real-IP $remote_addr;
      proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
      proxy_set_header Host $http_host;
      proxy_cache_bypass $http_upgrade;
      proxy_headers_hash_max_size 512;
      proxy_headers_hash_bucket_size 128;
     }
}