On Fri, 23 Nov 2001, Daniel Phillips wrote: > And ambitiously, under the heading of degrading gracefully: > > - try to delay the onset of thrashing by sheduling tasks with increasing > coarse granularity so that each running task is able to obtain a > reasonable working set of process memory and file cache. > yet under the degrading gracefully subject, the VM should clearly make the distinction between a valid high load and thrashing. The system should not start any load control/ process penalization if what we're having is a subtle (but perfectly valid) high load around. Formulating a little better (I guess): - there should be a clear definition of _what_ is a system degrading, so we can actually try to do it gracefully. -- Roberto -- -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ IRC Channel: irc.openprojects.net / #kernelnewbies Web Page: http://www.kernelnewbies.org/
