[Users] Beancounters, cgroup and writes

[Simon Boulet]

Hi Mark,

On Thu, Oct 17, 2013 at 5:02 PM, Mark J. <mjohanson at a2hosting.com> wrote:
> Noticing that the write counter is not incrementing in the beancounter file, nor is it allowing the cgroups to limit the writes either (guessing since the counter isn't changing the cgroup isn't being applied since its not seeing a change).
>
>

Current implementation in OpenVZ kernel only tracks synchronous
writes. It will not track asynchronous (because the OpenVZ kernel
doesn't have support for asynchronous page owner tracking). Reads are
always synchronous, that's why read throttling is working. Try
mounting the filesystem in synchronous mode and you should see blkio
throttle working as you expect.

I am not aware of any solution working under the OpenVZ kernel to do
asynchronous write throttling. Let me know if you find something that
does.

Simon