Queued for 3 hours is normal for this job type. Allow the job to run. https://wiki.galaxyproject.org/Support#Dataset_status_and_how_jobs_execute
That said, we believe that a few jobs may have a deployment problem to the cluster. So, even though in most cases we state to wait for queued jobs to execute for best performance - in this specific case, at this specific time, deleting the job and re-running is the recommendation for jobs that are queued for more than 10 hours. Wait for this or risk starting the queue process over again and extending the wait time even longer.
Our apologies for the confusion during this upgrade phase at http://usegalaxy.org.
Thanks, Jen, Galaxy team