Docker 1.12中的单机Swarm

Question

I would like to create docker-compose file in my development environment and use it to spin up a single machine "swarm". The goal would be to have the development environment be as consistent as possible with the CI, QA, and Prod environments.

I used docker a year+ ago but a lot has changed and I'm very interested in using 1.12 as my platform. My questions are as follows:

1. What is the difference between a "node" and a "physical machine"? Can a single machine (aka, a developer laptop) host multiple node's? My guess is that a node is virtual and that I should be able to have more than one but don't feel certain of it.
2. Assuming answer to #1 is that it is possible ... is there any reason these various nodes can't be "swarm workers" along with a singular "manager" all running on the laptop?

Note: I know it would be possible with VM's to emulate other machines -- many of the examples start off by doing this -- but I want to avoid running any VMs to lower the resource cost of running this setup

Are there any good examples of single-node swarms people can refer me to?

Answer 1

A node in the docker swarm is an instance of the docker engine configured in the swarm (with an init or join). An instance of a docker engine can only join up to a single swarm (so 0 or 1), so you can't create multiple nodes on the same engine. A typical developer install to test multiple nodes in a swarm is to spin up multiple VM's, each with a docker install.

You can have a swarm with a single manager which is also a worker. Tasks scheduled in a swarm may be scheduled on a manager just as they would a worker. Workers have no ability to manage the swarm, but managers have all the abilities of a worker. If you want to simply be able to run docker service commands, you can do a docker swarm init on yourself and then define your services there.