FYI, /adaql7/work1 (1.8TB!!) is 100% full. /adaql4/work1 (another 1.8TB!) is not far behind, nor is /adaql10/work1 (2.7TB). Fortunately, much of that usage does not seem to belong to the current experiment, but rather to "prex", "happex", "happexsp" and related collaborators, who (unsurprisingly) didn't clean up after themselves.
If DVCS runs low on disk space, you have my permission to remove old files from previous experiments/analyses. The run coordinator and Bob should coordinate.