Support request: What's wrong with my fig.yml for a wordpress?

[funkyfuture]

setting up a wordpress-instance with Docker is easy:

docker run --name="wp_db" -d -v /var/lib/mysql -e MYSQL_ROOT_PASSWORD=password mysql
docker run --name="wp_web" -d --link="wp_db:mysql" -v /var/www -p "8001:80" wordpress

to my knowledge, this would be the corresponding fig.yml:

web:
  image: wordpress
  links:
    - db:mysql
  ports:
    - "8001:80"
  volumes:
    - /var/www

db:
  image: mysql
  environment:
    MYSQL_ROOT_PASSWORD: password
  volumes:
    - /var/lib/mysql

however, starting the 'app' with fig up will lead to the situation that - to quote:

db_1  | 2015-01-18 20:55:44 1 [Warning] IP address '172.17.0.30' could not be resolved: Name or service not known
web_1 | MySQL Connection Error: (1130) Host '172.17.0.30' is not allowed to connect to this MySQL server
web_1 | 
web_1 | Warning: mysqli::mysqli(): (HY000/1130): Host '172.17.0.30' is not allowed to connect to this MySQL server in - on line 5

after inspecting the containers, this makes sense, since the failing containers are not attached to Docker's network-bridge. which i would expect since Docker does that by default.

anyway, also adding

  net: "bridge"

explicitly to both service's configuration will not create the network-bridge.

what's irritating even more is that it worked for a while with the fig.yml above. and the only aspect i changed these days was to update fig to version 1.0.1 (edit: with version 1.0.0 it's all the same.)

so, before i dig deeper i'd appreciate some feedback whether this is reproducable, a race condition to be expected or some obvious mistake on my side.

Ubuntu 14.04 / Mac OS X 10.9
Docker 1.4.1
fig 1.0.1

---

EDIT to elaborate a little more on the bridge-issue:

the verbose output states the last container start as follows:
docker start <- (u'62cd26fedcab626cdc212416cc81d753406542c8517e52415e58f0f7b6de44b5', links=[(u'wordpress_db_1', u'mysql'), (u'wordpress_db_1', u'wordpress_db_1'), (u'wordpress_db_1', u'db_1')], network_mode='bridge', binds={}, dns=None, volumes_from=[u'946af1607e985ba9eed6cc673ec45041928b14e1a2443ebc4e49c49e4ecf9700'], port_bindings={u'80': u'8001'}, privileged=False)

anyhow:

docker inspect 62cd26

(...)
    "NetworkSettings": {
        "Bridge": "",
        "Gateway": "",
        "IPAddress": "",
        "IPPrefixLen": 0,
        "MacAddress": "",
        "PortMapping": null,
        "Ports": null
    }

[funkyfuture]

today it worked reliable with Mac OS.
but within one minute i had all four possible, different outcomes on Ubuntu while i repeated fig rm && fig up. Either no container had a network-bridge, one of both had one or both had one. 😜

[funkyfuture]

after some further testing, i eventually figured out that it propably relates to to this issue.
it comes up when there are one ore more VPNs used on the Docker-host. however it still isn't reliably reproducable.