Unable to find Jank

[Don Garrett]

Jay and I (well, mostly Jay) been trying to reproduce/measure UI performance jank when update_engine runs in background. So far, we haven't been able to do it. Is there anyone that would like to help out?

If it helps our best guess candidate for a problem cause is that ionice isn't being very effective. Our second best guess is the additional memory pressure, though what we use (other than io buffers) is mostly measured in low 10's of megs.

--
Don

[Mandeep Singh Baines]

Don Garrett (dgar...@chromium.org) wrote:
> Jay and I (well, mostly Jay) been trying to reproduce/measure UI
> performance jank when update_engine runs in background. So far, we haven't
> been able to do it. Is there anyone that would like to help out?
>

Is there a bug number?

> If it helps our best guess candidate for a problem cause is that ionice
> isn't being very effective. Our second best guess is the additional memory
> pressure, though what we use (other than io buffers) is mostly measured in
> low 10's of megs.
>

I find mpstat and vmstat to be great tools for detecting jank I've heard
that latency_top is also a good tool though I haven't used it much.

Regards,
Mandeep

> --
> Don
>
> --
> Chromium OS Developers mailing list: chromiu...@chromium.org
> View archives, change email options, or unsubscribe:
> http://groups.google.com/a/chromium.org/group/chromium-os-dev?hl=en

[Don Garrett]

Just to point out to everyone. We haven't been able to reproduce this when we're watching, but there have been a lot of reports from canary channel and such. The bug is chromium-os:27597.

--
Don

[Don Garrett]

Um... chromium-os:27957

--
Don