actually this is a admin problem. Are you running Galaxy with a queueing system? Galaxy supports several scheduling systems that will manage your jobs and create Queues. If you have to many jobs on one cluster, like in your case, the Blast Jobs have to wait until your cufflinks job is finished.
Please tell your Galaxy admin to look use such a scheduler and you will be fine :)
Your question is a bit difficult to understand.
If you are talking about a local or cloud Galaxy with an associated cluster/set of nodes actually executing the jobs, then you are probably seeing the resources used by Galaxy based on interactions with the database itself.
Best, Jen, Galaxy team