Every container is accessible through its own IP address. For instance, we can find the IP address of our elasticsearch-oss container by running docker inspect, and looking under NetworkSettings.IPAddress:

$ docker inspect a415f4b646e3
[
    {
        "Id": "a415f4b646e3a71...81b9b7f6dcaf76b",
        "Created": "2018-05-10T19:37:55.565685206Z",
        "Image": "sha256:3822ba554fe9...adf11f6a59bde0139",
        "Name": "/elasticsearch",
        "Driver": "overlay2",
        "NetworkSettings": {
            "Ports": {
                "9200/tcp": [{
                    "HostIp": "0.0.0.0",
                    "HostPort": "9200"
                }],
                "9300/tcp": [{
                    "HostIp": "0.0.0.0",
                    "HostPort": "9300"
                }]
            },
            "Gateway": "172.17.0.1",
            "IPAddress": "172.17.0.2",
            ...
        }
        ...
    }
]

You can also use the --format or -f flag to retrieve only the field you are interested in:

$ docker inspect -f '{{.NetworkSettings.IPAddress}}' elasticsearch
172.17.0.2

However, our local instance of our API assumes that Elasticsearch is available on localhost:9200, not 172.17.0.2. If we are going to provide an equivalent behavior to our non-containerized Elasticsearch, we must make Elasticsearch available on localhost:9200. That’s the job of the -p flag.

The -p flag publishes a port of the container and binds it to the host port:

$ docker run -p <host-port>:<container-port>

In our case, we are binding the 9200 port of 0.0.0.0 to the 9200 port of the container. 0.0.0.0 is a special address that refers to your local development machine.