Logs are not available through Docker API when running on AWS

[markbirbeck]

When working on #17 I discovered that something quirky happens at the point where we try to retrieve the logs:

$ dj --showlogs --ssh-hostname docker@ec2-[snip].amazonaws.com --ssh-identity-file ../docker-swarm.pem --ssh-remote /var/run/docker.sock hello-world
6jrmk4pm[snip]m27rt
About to get logs for container: e5afa446e1fa9c233c2[snip]542bba66593374c53085fd0719cdc42a4
kgeh0mfldrdpz85lc8lgx8m0e: (HTTP code 404) unexpected - No such container: e5afa446e1fa9c233c2[snip]542bba66593374c53085fd0719cdc42a4

Although the error message is a bit strange, it's almost certainly caused by AWS sending all logs to CloudWatch rather than Docker looking after them itself.

I'm not sure if it's possible to tell the Swarm to use a different log driver when running the service, or whether AWS just ignores those settings. It might mean trying to get the logs from CloudWatch which could be a right pain.

[markbirbeck]

It might be best to move this issue down to the docker-engine module.

[markbirbeck]

Looks like it is possible to change the logging policy at the container level, at least, with something like:

docker -H localhost:2374 run --log-driver json-file hello-world

So we'll have to play around and see if we can do the same thing with a service.