Ralf Sigmund wrote: > Bernard Li wrote: > >> Hi Malay: >> >> Are there any documentations and/or papers which describe such a setup? >> I would assume that there would be general interest in seeing how such a >> setup could be implemented. >> I really don't know what you meant my this. If you mean splitting the input jobs then I do most of my work on LSF, and I use my own custom script to do this. If more people are interested I can port it to other systems. The script is very intelligent in that it can detect machine faulire, NSF problem and create a black-listed node before relauching jobs. If you meant "pull-technique" then though I worked on it for some-time, I gave up because of machine access issues. The method require more access to nodes which in normal setup scenario simply not available for a normal user. As I consider primarily a user, I simply don't like working on anything that's not available for normal users. Having said that, there are ways and means to actually set it up by a normal user (my abandoned work might give some clues). If any one of you interested I might revive the project. -Malay mbasu(at)ncbi.nlm.nih.gov