Nice catch. Yes, I think the graphs show that we have more/longer hangs in the 1-30 second range on M53 beta than we used to have on M52 beta.

There is a similar bump on Linux (but not other OS), which might be coming back down now. Maybe once CrOS revs again it could become normal?

rtenneti@ are you actively investigating this one? This is currently a stable blocker.

Hi  ketakid@,
  We in chrome networking noticed that IO thread became janky. In Canary and Dev, we had seen this regression around July 7th time frame in Canary and July 28th in Dev.

+ mmenke@, jkarlin@chromium.org and csharrison@

https://uma.googleplex.com/p/chrome/timeline_v2/?sid=024c1ac07c29a93136b7cb070c283d9d

Would appreciate if someone could dig deeper. Thanks.

Hrm...The regression I'm looking at is M54, not M53, but very clearly shows up on ThreadWatcher.ResponseTime.IO on Windows https://uma.googleplex.com/p/chrome/timeline_v2?sid=d9a0e0b4064b51f0f1ee7bb77ce37208 (Less clear if there's something going on over on the UI thread).  ChromeOS is also pretty weird, maybe because of sparse data:  https://uma.googleplex.com/p/chrome/timeline_v2?sid=dc7948ffa3a239826f9875714998f893

rtenneti@ Do we see this janky behavior on the latest beta M53?

No, the regression he's describing happened in M54.

Just to clarify, beta M53 has a pretty clear regression on these values (Which is seen in the latest version of the beta), while M54 looks to have a much bigger regression, and that one (issue 627999) is being investigated by me, off and on.

The behavior on M53 looks like something that needs to investigated asap as well. WE are about to hit Stable end of month and this bug is currently marked a blocker for that. mmenke@ do you know who can investigate this issue on M53 branch?

No, I don't.  This issue isn't clearly a problem with any particular component, unfortunately.  Something weird is happening somewhere on the IO thread...what that is, or what's doing it, is completely unclear.

We are getting close to Stable next week. We need a call on whether we are planning on investigating or punting this to the next milestone.

Moving this bug to M54.

Is this still active? I can't repro on my Samus running latest M54

The M53 regression this bug was filed for does not look to have been fixed, looking at the numbers.

The M54 cross-platform (At least Win/OSX) ThreadWatcher.ResponseTime.IO regression I mentioned above looks to be fixed.

There also looks to be a new bonus regression in ThreadWatcher.ResponseTime.UI in M55.  Exciting times!

This issue has been Available for over a year. If it's no longer important or seems unlikely to be fixed, please consider closing it out. If it is important, please re-triage the issue.

Sorry for the inconvenience if the bug really should have been left as Available. If you change it back, also remove the "Hotlist-Recharge-Cold" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot