search

radial / wheel-coreos-pxe

Wheel repository for running dnsmasq as PXE+DHCP server serving Coreos images.
https://registry.hub.docker.com/u/radial/coreos-pxe/
26 stars 9 forks source link

dynamic launching #1

Open hh opened 9 years ago

hh commented 9 years ago

It's very likely my not understanding everything, and since you are actively working on it I figured I make sure I'm taking the right approach.

I have a fig-dynamic.yml:

logs:
    image: "radial/axle-base:latest"
    volumes:
        - /log
hub:
    image: "radial/hub-base:latest"
    volumes_from:
        - logs
    volumes:
        - /config
        - /data
        - /run
    environment:
        WHEEL_REPO_PXE: "https://github.com/hh/wheel-coreos-pxe.git"
pxe:
    image: "radial/coreos-pxe:latest"
    volumes_from:
        - hub
    net: "host"

Which should be grabbing from: https://github.com/hh/wheel-coreos-pxe/compare/config

$ fig -f ./fig-dynamic.yml  up
Recreating wheelcoreospxe_logs_1...
Recreating wheelcoreospxe_hub_1...
Recreating wheelcoreospxe_pxe_1...
Attaching to wheelcoreospxe_logs_1, wheelcoreospxe_pxe_1
logs_1 | Container will now idle.
pxe_1  | Spoke "pxe" is waiting for Hub container to load...

And it just hangs.

$ docker ps
CONTAINER ID        IMAGE                      COMMAND                CREATED             STATUS              PORTS               NAMES
5eee8099e711        radial/coreos-pxe:latest   "/spoke-entrypoint.s   21 seconds ago      Up 20 seconds                           wheelcoreospxe_pxe_1
dda43fb1312f        radial/axle-base:latest    "IDLE"                 23 seconds ago      Up 22 seconds                           wheelcoreospxe_logs_1

I don't see it launching a hub_1

Any thoughts?

brianclements commented 9 years ago

Did you try doing a complete restart from scratch? Do fig stop && fig rm to have it delete all the containers and rebuild it from nothing. I'm weary of figs ability to properly "recreate" when using fig up over and over. Also, when I've had situations like this, containers not starting at all, it was usually some type of problem with fig under the hood and what seemed like a race condition somewhere.