Restarting Upstart Services Doesn't Re-read the config about garethr-docker HOT 2 CLOSED

This is less of a problem of Upstart, and more one of Docker itself.

Because open ports can only be specified with docker run, and cannot be changed afterwards (one could argue that this is "by design" for Docker), changing the configuration of a container requires removing the container, and start a new container.
Just docker restarting a container does not suffice.

The problem with this is that the container might contain data the user wants to keep, but which would be lost when doing docker rm on the container. Some container can safely be deleted and rerun (consider a Skydock container that contains an immutable configuraiton), other containers should only be removed if the user explicitly requests to do so.

Proposed fix: Add a parameter to docker::run which allows the user to specify whether containers should be temporary (i.e. removed on initctl stop) or permanent. Would it be possible to use --rm parameter of docker run. Additionally, one could define an additional action for the init scripts to explicitly remove and rerun the container (Is this possible for Upstart?).

Update: According to the the Upstart Documentation calling initctl restart <jobname> does not reload a jobs config from disk.

from garethr-docker.

We've now changed over to sysv from docker run which should avoid this issue and a few other init related problems.

from garethr-docker.