Hi Ivo, Ivo Grosse wrote: > > Hi Jim, > > > A cluster of client workstations, using the clients idle cycles, works > > great until your thesis / CFO's budget / experimental results get > > destroyed by an errant batch process. > > Can you please clarify this??? > > It's clear that I would not submit my "latex" or "emacs" or "xmgrace" > job to the queue. No question, so not that scenario. > Do you mean that those non-batched jobs could be > killed by batched jobs? Possibly killed, (possibly run out of /tmp space or disk space in general, benign DOS attack, etc.), and it depends on what kind of risk you are willing to take with your clients data while the batch process is sharing the machine with them and vice versa. Or to put it another way, who in your group is willing to have a batch process run the background on their systems idle cycles while they are working on their own projects? If you have racks of unused old Sun's and SGI's and want to make a new cluster without clients I think this is a different problem. Jim > Ivo > > _______________________________________________ > Bioclusters maillist - Bioclusters@bioinformatics.org > http://bioinformatics.org/mailman/listinfo/bioclusters