Memory limits on Helium's all.q queue

In order to make the most efficient use of limited resources there are memory limits in place for jobs that share nodes on Helium's sandbox, all.q and UI queues. The limits are set to be approximately equal to the amount of physical memory and if a job hits the limit it will be aborted. During the maintenance of June 6, 2014, we will be expanding the use of memory limits to include all jobs, even those that request whole nodes, running in the all.q queue. This is being done to mitigate the risk of jobs in the all.q queue rendering a machine unavailable to the corresponding investor queue due to memory oversubscription.