Implement logging for rancher containers

[solocommand]

Use logstash or some other log storage/aggregation service and load logspout on all rancher hosts to store container logs. Ensure that they are stored and can be used for troubleshooting purposes.

Currently when rancher kills and restarts a container there is no way to retrieve the logs for said container to diagnose what caused it to become unhealthy.

We need this information both for reactive diagnoses like this, as well as to spot future problems within the container infrastructure.

It'd be a good idea to also route syslog from all ec2 instances to this service as well to spot infrastructure level issues, in addition to the docker container logging.

[zarathustra323]

@solocommand @Jlaird Status on this?