Dockerize?

[prologic]

Could we Dockerize this and create an appropriate Dockefile with relevant how-to on how to spin it up in the README? The container that runs docker-sync would have to run privileged? Or with a link to the host docker daemon via -v/--volume?

[prologic]

If all this is possible (which I think it is) the resulting image should then be published to the Docker Public Registry :)

[blalor]

Yeah, I think that is possible. Not sure about how useful, but possible. :-) It wouldn't even have to run privileged, but it would have to have access to the Docker socket.

[prologic]

IHMO I think it would be very useful because
you could "create your docker infrastructure"
based on a "configuration" and just spin up
"docker-sync".

James Mills / prologic

E: prologic@shortcircuit.net.au
W: prologic.shortcircuit.net.au

On Fri, Apr 4, 2014 at 9:19 PM, Brian Lalor notifications@github.comwrote:

Yeah, I think that is possible. Not sure about how useful, but
possible. :-) It wouldn't even have to run privileged, but it would
have to have access to the Docker socket.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/1#issuecomment-39554796
.

[blalor]

There's an interesting chicken/egg problem here. Since docker-sync manages the entire set of containers in the Docker instance it's pointing to, it will actually terminate itself if there's no associated config file that exactly matches the container that's running docker-sync. So you'd need a (basically no-op) container config for docker-sync itself included alongside your other container configs, like:

---
name: docker-sync
image: autosportlabs/docker-sync:latest

volumes:
    /var/run/docker.sock:
        HostPath: /var/run/docker.sock

and then run the docker-sync image in such a way that this config matches exactly with the container that's running docker-sync.

docker run \
    -v /var/run/docker.sock:/var/run/docker.sock \
    --name=docker-sync \
    autosportlabs/docker-sync:latest

However, since you can only have one container with a given name, you have to docker rm docker-sync before you can sync again.

I suppose the alternative would be to add an option to not purge unmanaged containers, and I've been thinking that would be useful in some cases (such as non-long-running containers).

[prologic]

Precisely :) But with any chick/egg
problem you can solve it by creating
a fake or half/chicken before hand :)

cheers
James

James Mills / prologic

E: prologic@shortcircuit.net.au
W: prologic.shortcircuit.net.au

On Fri, Apr 4, 2014 at 10:30 PM, Brian Lalor notifications@github.comwrote:

There's an interesting chicken/egg problem here. Since docker-sync manages
the entire set of containers in the Docker instance it's pointing to, it
will actually terminate itself if there's no associated config file that
exactly matches the container that's running docker-sync. So you'd need a
(basically no-op) container config for docker-sync itself included
alongside your other container configs, like:

---name: docker-syncimage: autosportlabs/docker-sync:latest
volumes:
/var/run/docker.sock:
HostPath: /var/run/docker.sock

and then run the docker-sync image in such a way that this config matches
exactly with the container that's running docker-sync.

docker run -v /var/run/docker.sock:/var/run/docker.sock --name=docker-sync autosportlabs/docker-sync:latest

However, since you can only have one container with a given name, you have
to docker rm docker-sync before you can sync again.

I suppose the alternative would be to add an option to not purge unmanaged
containers, and I've been thinking that would be useful in some cases (such
as non-long-running containers).

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/1#issuecomment-39559533
.