Unable to reach web server in Docker swarm from th

I'm starting out using Docker on macOS, and get stuck when trying to complete part 4 of the Get Started guide. I have created two extra virtual machines (myvm1 and myvm2), set myvm1 as swarm manager, and myvm2 as a worker. I have then deployed a stack with 5 Flask web servers using the docker-compose.yml from part 3 of the tutorial. The processes seem to start fine, and are distributed between the two machines, but I am not able to reach them from the host using a browser.

How should I configure the port forwarding/network to be able to reach the web servers in the swarm from the host of the virtual machines running the docker container?

The following is a list of commands that I have run, some with resulting output.

$ docker-machine create --driver virtualbox myvm1
$ docker-machine create --driver virtualbox myvm2

$ docker-machine ls
NAME    ACTIVE   DRIVER       STATE     URL                         SWARM   DOCKER     ERRORS
myvm1   -        virtualbox   Running   tcp://192.168.99.100:2376           v18.09.0
myvm2   -        virtualbox   Running   tcp://192.168.99.101:2376           v18.09.0

$ docker-machine ssh myvm1 "docker swarm init --advertise-addr 192.168.99.100"
$ docker-machine ssh myvm2 "docker swarm join --token <my-token-inserted-here> 192.168.99.100:2377"

$ eval $(docker-machine env myvm1)
$ docker-machine ls
NAME    ACTIVE   DRIVER       STATE     URL                         SWARM   DOCKER     ERRORS
myvm1   *        virtualbox   Running   tcp://192.168.99.100:2376           v18.09.0
myvm2   -        virtualbox   Running   tcp://192.168.99.101:2376           v18.09.0

$ docker stack deploy -c docker-compose.yml getstartedlab

$ docker stack ps getstartedlab
ID                  NAME                  IMAGE                            NODE                DESIRED STATE       CURRENT STATE              ERROR               PORTS
it9asz4zpdmi        getstartedlab_web.1   mochr/test_repo:friendly_hello   myvm2               Running             Preparing 18 seconds ago
645gvtnde7zz        getstartedlab_web.2   mochr/test_repo:friendly_hello   myvm1               Running             Preparing 18 seconds ago
fpq6cvcf3e0e        getstartedlab_web.3   mochr/test_repo:friendly_hello   myvm2               Running             Preparing 18 seconds ago
plkpximnpobf        getstartedlab_web.4   mochr/test_repo:friendly_hello   myvm1               Running             Preparing 18 seconds ago
gr2p8a0asatb        getstartedlab_web.5   mochr/test_repo:friendly_hello   myvm2               Running             Preparing 18 seconds ago

The docker-compose.yml:

version: "3"
services:
  web:
    image: mochr/test_repo:friendly_hello
    deploy:
      replicas: 5
      resources:
        limits:
          cpus: "0.1"
          memory: 50M
      restart_policy:
        condition: on-failure
    ports:
      - "4000:80"
    networks:
      - webnet
networks:
  webnet:

It looks like this is a known problem with the current version of boot2docker: https://github.com/docker/machine/issues/4608

The workaround is either to use a swarm based on machines that do not require boot2docker (e.g. AWS, DigitalOcean, etc.), wait until a newer version of boot2docker is released, or use an earlier version of boot2docker, as described in that link. To use an earlier version:

export VIRTUALBOX_BOOT2DOCKER_URL=https://github.com/boot2docker/boot2docker/releases/download/v18.06.1-ce/boot2docker.iso

before creating your virtual machines with docker-machine. (Remove your existing virtual machines first, then use that export, then docker-machine create myvm1)

Then, you should be able to bring up your stack and access your containers at either 192.168.99.100:4000 or 192.168.99.101:4000 (or whatever IP addresses are revealed by docker-machine ls)