I guess you can develop a new activity, say, with the name "job scheduler". When the workflow executes, this activity will forward the job description and data to a job scheduler which can distribute the jobs in a cluster.
-Wei Fan Dong wrote: > > Hi, > > I have long been puzzled by this – how exactly can we send a workflow > to a cluster (say SGE) such that heavy jobs included in that workflow > can be distributed onto different nodes inside the cluster? > > I noticed that the workflow can be executed outside Taverna using > command lines executeworkflow.sh (is that true for Taverna 2.X? Seems > I cannot find this shell script in Taverna 2.1). If I run a workflow > in this way from one of the nodes in the cluster, how should I do to > distribute that workflow to different nodes? Not sure if I make myself > clear enough, from my previous experience working on the cluster, I > usually cut a big job into small pieces which are distributed by SGE > queue. At the end, I use another script to merge results returned from > different nodes. But I cannot think of a way to use that practice on a > workflow. > > Any inputs would be highly appreciated. > > Fan > > ------------------------------------------------------------------------ > > ------------------------------------------------------------------------------ > > ------------------------------------------------------------------------ > > _______________________________________________ > taverna-users mailing list > [email protected] > [email protected] > Web site: http://www.taverna.org.uk > Mailing lists: http://www.taverna.org.uk/taverna-mailing-lists/ > -- Wei Tan, Ph.D. Computation Institute the University of Chicago|Argonne National Laboratory http://www.mcs.anl.gov/~wtan ------------------------------------------------------------------------------ _______________________________________________ taverna-users mailing list [email protected] [email protected] Web site: http://www.taverna.org.uk Mailing lists: http://www.taverna.org.uk/taverna-mailing-lists/
