On Tue, 27 Jan 2009, Evgeniy Polyakov wrote: > Having some special application which will monitor /dev/mem_notify and > kill processes based on its own hueristics is a good idea, but when it > fails to do its work (or does not exist) system has to have ability to > make a progress and invoke a main oom-killer. > Agreed, very similiar to the cgroup oom notifier patch that invokes the oom killer if there are no attached tasks waiting to handle the situation. In this case, it would be a configurable delay to allow userspace to act in response to oom conditions before invoking the kernel oom killer. So instead of thinking of this as a userspace replacement for the oom killer, it simply preempts it if userspace can provide more memory, including the possibility of killing tasks itself. _______________________________________________ Containers mailing list Containers@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/containers