Mike Galbraith wrote: > On Sat, 2006-06-03 at 10:08 +1000, Peter Williams wrote: >> Mike Galbraith wrote: >>> How? How would you deal with the make example with per task caps. >> I'd build a resource management tool that uses task statistics, nice and >> caps to manage CPU resource allocation. This could be a plug in kernel >> module or a user space daemon. It doesn't need to be in the scheduler. > > Ok, you _can_ gather statistics, and modify caps/nice on the fly... for > long running tasks. How long does a task have to exist before you have > statistics for it so you can manage it?
If the stats package is up to scratch it will provide stats for tasks that have exited so you will be able to charge their resource usage to the higher level entity and still manage that entity's usage properly via its other tasks. > > Also, if you're going to need a separate resource manager to allocate, > monitor and modify in realtime, why not go whole hog, and allocate and > monitor instances of uml. It'd be a heck of a lot easier. Or Xen. Or Vmware. That would be one solution and brings other functionality that may be desirable. Of course, you can also do resource control within those instances as well :-). "There's more than one way to skin a cat" as the old saying goes. Peter -- Peter Williams [EMAIL PROTECTED] "Learning, n. The kind of ignorance distinguishing the studious." -- Ambrose Bierce _______________________________________________ ckrm-tech mailing list https://lists.sourceforge.net/lists/listinfo/ckrm-tech
