Article ID: 1459 - Last Modified: April 20, 2011
If I need to kill a Desmond FEP or Replica Exchange job, how do I make sure that I get the results back?
Killing a Desmond job that involves subjobs can result in the loss of information from partially completed subjobs. For instance, for an FEP job there are typically 12 lambda windows, each of which is an independent subjob that runs a Desmond simulation. In the main sampling stage (the most CPU intensive stage) you may have some subjobs that are complete and some that are still running when a job is killed. The results for completed stages and completed subjobs should be transferred back successfully to the launch directory. However, the files for the partially completed subjobs won't be. This can be a problem if, for example, the subjobs are nearly done. One way to avoid this problem is to copy the files from the scratch directory for the subjob just before killing the overall job.
Type the words or phrases on which you would like to search, or click here to view a list of all
Knowledge Base articles