I think we need to add a configuration to the global lxc.conf whereas any given container may run only on one NUMA node, and if that is not possible, it should not even start.
The performance for a container that is contained, so to speak, in a single NUMA node, should be much higher that a container that has its processes all over the place. I noticed this in Hyper-V virtual machines, where you may set this restriction. The performance as measured by hdparm tT --direct is twice than a virtual machine started without this feature. This is the distance between nodes, as reported by numactl --hardware node distances: node 0 1 2 3 0: 10 20 20 20 1: 20 10 20 20 2: 20 20 10 20 3: 20 20 20 10 _______________________________________________ lxc-users mailing list [email protected] http://lists.linuxcontainers.org/listinfo/lxc-users
