Currently, the Marathon plugin will issue a deployment and succeed if the deployment is created. But it does not verify that the deployment completes or is successful. In a common CI/CD scenario, where a bad development build leads to a bad deployment, there's no feedback from the plugin that a failure has occurred, and thus the job will succeed, but the deployment will be left in e.g. Mesos retrying forever until someone notices it. 

The Marathon plugin should monitor the deployment for a configurable period and fail if it does not succeed, rather than simply firing it off and assuming it was successful.