Messages in this thread | | | Date | Wed, 30 May 2012 17:44:43 -0700 (PDT) | From | David Rientjes <> | Subject | Re: [PATCH] meminfo: show /proc/meminfo base on container's memcg |
| |
On Thu, 31 May 2012, Kamezawa Hiroyuki wrote:
> My test with sysfs node's meminfo seems to work... > > [root@rx100-1 qqm]# mount --bind /sys/devices/system/node/node0/meminfo > /proc/meminfo > [root@rx100-1 qqm]# cat /proc/meminfo > > Node 0 MemTotal: 8379636 kB
This doesn't seem like a good idea unless the application supports the "Node 0" prefix in /proc/meminfo.
If any application really cares about the amount of memory available to it, it should be taught to be memcg aware. Then do something like
cat $(grep memory /proc/mounts | cut -d " " -f 2)/$(grep memory /proc/self/cgroup | cut -d : -f 3)/memory.limit_in_bytes
but since that value can change all the time then it doesn't seem helpful unless we have a userspace notifier.
| |