Mulyadi Santosa wrote: > Hi Li... > > On Wed, Jun 18, 2008 at 10:09 AM, Li Zefan <lizf@xxxxxxxxxxxxxx> wrote: >> Yes, it's something like CKRM. Some years ago we had several proposals like CKRM, >> ResGroups, User BeanCounters and others. > > Interesting....I'll read more about it ASAP. I'm kinda interested in > resource management lately. > > But if I may know, using current state of cgroup, can I really do fine > grained resource control? I mean, if I "trim" certain process group > CPU usage not more than 20% of total CPU time, is it really 20%? or if > there's "tolerance", how big is it according to real life test? > That's cpu subsystem - fair group scheduler. You can try it with a new kernel: # mount -t cgroup -o cpu xxx /mnt But it's soft limit not hard limit, I mean, if you limit a group's cpu time to 20%, it can get 100% if no other processes in other groups are running. And this cpu subsystem is still far from perfect, Peter Zijlstra is working on it. -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
