elasticipd

As it is now common practice to run applications on top of a container-orchestration platforms, such as Kubernetes, there is no guarantee that a service will always run on the same host. This can cause problems when a service requiring a public IP address gets rescheduled. elasticipd automatically associates an Elastic IP address to the AWS EC2 instance running this service. It is designed to run as a sidecar container alongisde a service that requires a public IP address.

Usage

As elasticipd is currently configured to use AWS Instance Roles, the host will need to have and EC2 Policy with at least the following actions: DescribeInstances, DescribeInstanceAttribute, AssociateAddress and DisassociateAddress.

The service is configured by setting the following command line flags:

Usage of elasticipd:
  -elastic-ip string
        Elastic IP address to associate
  -interval string
        Attempt association every interval (default "30s")
  -port int
        Local HTTP server port (default 8081)
  -reassoc
        Allow Elastic IP to be reassociated without failure (default true)
  -region string
        AWS region hosting the Elastic IP and EC2 instance
  -retries int
        Maximum number of association retries before fatally exiting (default 3)

Kubernetes

A simple multi-container Pod spec:

apiVersion: v1
kind: Pod
metadata:
  name: public-service
spec:
  containers:
  - name: public-service
    image: public-service
  - name: elasticipd
    image: ghcr.io/loshz/elasticipd:v2.3.0
    command: ["elasticipd"]
    args: [
        "-elastic-ip=<elastic_ip>",
        "-region=us-west-2",
        "-interval=10s"
    ]
    ports:
    - containerPort: 8081
      name: local-http
    livenessProbe:
      httpGet:
        path: /healthz
        port: local-http
      initialDelaySeconds: 3
      periodSeconds: 3