Thanks for the reviews, I'll merge this series shortly.
Ethan
On Mon, Apr 1, 2013 at 3:54 PM, Ben Pfaff wrote:
> On Mon, Apr 01, 2013 at 03:50:22PM -0700, Ethan Jackson wrote:
>> > Currently, I think, we push stats at least once a second. This
>> > reduces that to at least once every two second
On Mon, Apr 01, 2013 at 03:50:22PM -0700, Ethan Jackson wrote:
> > Currently, I think, we push stats at least once a second. This
> > reduces that to at least once every two seconds, I think. Does
> > pushing once a second reduce performance?
>
> It basically wipes out the benefit entirely. If
> Currently, I think, we push stats at least once a second. This
> reduces that to at least once every two seconds, I think. Does
> pushing once a second reduce performance?
It basically wipes out the benefit entirely. If you feel that this
approach is risky, I'd prefer to skip the patch entire
On Sun, Mar 31, 2013 at 06:22:58PM -0700, Ethan Jackson wrote:
> The most natural place to push facet statistics is in
> update_stats() where they're pulled from the datapath. However,
> under load, update_stats() can be called as many as 10 times per
> second causing us to push statistics so freq
