LSF cluster

[khayer]

Not all LSF clusters have plus queue.

[delagoya]

Same could be said for the SGE command line. Specifically not all SGE clusters have a make queue, and, more importantly, if that queue may not have the $pe_slots allocation rule. In the case of $fill_up would make a best effort to schedule the parallel jobs onto the same node, but they may end up on different nodes, which would do two things:

1. it would run the job twice
2. It has the possibility of overloading the execution node's resources, since one of the threads would not be accounted for.

[khayer]

Good point. For now --platform-flags can be over-written, but this is certainly not the ideal solution.