consul-agent.json5

/*
This example demonstrates how a user can create a chain of dependencies
on a consul-agent process that runs inside the container (as is required
in CaaS environments like sdc-docker).

'app' is a Node.js application that requires configuration from
Consul before it can start. The 'preStart' job that configures 'app'
needs 'consul-agent' to be healthy before it can run.
*/
{
  consul: "localhost:8500", // ContainerPilot will talk to the local agent
  jobs: [
    {
      // this job has no 'port' configuration so it will not be
      // advertised to the Consul server. note there's no 'when' field
      // so this will start on the 'global startup' event by default.
      name: "consul-agent",
      exec: [
        "consul", "agent",
        "-rejoin",
        "-retry-join", "{{ .CONSUL }}",
        "-retry-max", "10",
        "-retry-interval", "10s"
      ],
      restarts: "unlimited",
      health: {
        // this health check will result in 'consul-agent healthy' events
        // that can be consumed by the 'when' field of our 'preStart'
        exec: "consul info | grep leader"
      }
    },
    {
      // this job is not advertised and has no health check. we'll never
      // see a 'preStart healthy' event, just 'preStart exitSuccess'
      name: "preStart",
      exec: [
        "consul-template", "-once", "-consul-addr", "localhost:8500",
        "-template", "/etc/template.ctmpl:/etc/app.conf"
      ],
      restarts: "never", // default behavior
      when: {
        // we'll wait for the "consul-agent healthy" event, which appears
        // inside this container only as it's for a non-advertised job.
        // note we don't need to script a poll+timeout.
        source: "consul-agent",
        once: "healthy",
        timeout: "120s"
      }
    }
    {
      name: "app",
      exec: [
        "/usr/local/bin/node",
        "/usr/local/bin/http-server", "/srv", "-p", "8000"],
      port: 8000, // this job will be advertised to Consul servers
      restarts: "unlimited",
      when: {
        // 'app' won't start until the 'preStart' has succeeeded, but we
        // give up after 120 seconds
        source: "preStart",
        once: "exitSuccess"
        timeout: "120s"
      },
      health: {
        exec: "/usr/bin/curl --fail -s -o /dev/null http://localhost:8000",
        interval: 5,
        ttl: 10
        timeout: "10s" // the health check can have its own timeout
      }
    }
  ]
}