On 30.11.2020 19:40, Peter Xu wrote:
On Mon, Nov 30, 2020 at 11:11:00AM +0300, Andrey Gruzdev wrote:
On 28.11.2020 01:28, Peter Xu wrote:
On Thu, Nov 26, 2020 at 06:17:34PM +0300, Andrey Gruzdev wrote:
Since reading UFFD events and saving paged data are performed
from the same thread, write fault latencies are sensitive to
migration stream stalls. Limiting total page saving rate is a
method to reduce amount of noticiable fault resolution latencies.
Migration bandwidth limiting is achieved via noticing cases of
out-of-threshold write fault latencies and temporarily disabling
(strictly speaking, severely throttling) saving non-faulting pages.
So have you done any measurements out of it, as we've talked in previous
version? Thanks,
Sorry, not done yet.
So do you still plan to? :)
And if not, could you describe the rational behind this patch? For example,
what's the problem behind (e.g., guest hangs for xxx seconds, maybe?) and
what's the outcome (guest doesn't hang any more)?
Thanks,
Yes, I think providing a latency histogram with description of the test
case is the right thing.
--
Andrey Gruzdev, Principal Engineer
Virtuozzo GmbH +7-903-247-6397
virtuzzo.com
