loshz / ingressd

Automatic AWS Route53 IP address association for ingress services

NOTE: Use a load balancer instead.

ingressd

A lightweight daemon used to update Route53 records with the IP addresses of your ingress services, as well as perform health checks on desired hosts.

Architecture

0. Configure ingressd with list of Route53 host records.
1. Query EC2 for nodes with a specific tag, and return their public IP addresses.
2. Make several health checks against each ingress service IP address with specific host header (curl -H "Host: example.com" http://192.168.0.1).
3. Update Route53 records with IP addresses that have passed all health checks.

Usage

As ingressd is currently configured to use AWS Instance Roles, the host will need to have a role with at least AmazonEC2ReadOnlyAccess and a Route53 policy with the following actions:ChangeResourceRecordSets, ListResourceRecordSets, ListHostedZones.

Config

The service can be configured by setting the following environment variables:

Name	Type	Description
AWS_EC2_TAG	string	key:value of EC2 tag to query for instances
AWS_REGION	string	AWS region of EC2 instances to query
AWS_ROUTE53_RECORDS	string slice	Comma separated list of Route53 records to be updated
POLL_INTERVAL	string	Poll interval for Route53 updates
PORT	int	Port to bnd the local HTTP server to

Kubernetes

A simple single container Pod spec:

apiVersion: v1
kind: Pod
metadata:
  name: ingressd
  labels:
    app.kubernetes.io/name: ingressd
spec:
  securityContext:
    runAsUser: 2000
    runAsGroup: 2000
    fsGroup: 2000
  containers:
  - name: ingressd
    image: quay.io/syscll/ingressd:v0.1.0
    command:
    - ingressd
    livenessProbe:
      httpGet:
        path: /healthz
        port: 8081
      initialDelaySeconds: 5
      periodSeconds: 3
    ports:
    - containerPort: 8081
    env:
    - name: AWS_EC2_TAG
      value: "Name:haproxy"
    - name: AWS_REGION
      value: "eu-west-1"
    - name: AWS_ROUTE53_RECORDS
      value: "syscll.org,ingress.syscll.org,haproxy.syscll.org"
    - name: POLL_INTERVAL
      value: "10s"

TODO

• Allow host health check configuration