New issue
Advanced search Search tips

Issue 781013 link

Starred by 1 user

Issue metadata

Status: WontFix
Owner: ----
Closed: Nov 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Bug



Sign in to add a comment

Implement Navigation.TimeToNetworkStack UMA

Project Member Reported by csharrison@chromium.org, Nov 2 2017

Issue description

I've noticed a trend where navigation throttles observe WillStartRequest, and post a task to the IO thread to do some async work.

This IO thread work blocks navigation!!! I.e. since the IO thread is FIFO this work will occur before NavigationURLLoaderImplCore::Start.

In fact, I think  issue 779914  might be in this critical path.

I firmly believe adding this metric will prevent PLT regressions moving forward, as it will be extremely low noise, and 100% in the critical path of loading (if restricted to main frames).
 
Note this also applies to the (few) observers of  DidStartNavigationToPendingEntry, which includes the net predictor.
Oh wait, is this Navigation.TimeToURLJobStart?
Cc: carlosk@chromium.org
Status: WontFix (was: Untriaged)
yupp, thanks carlosk for doing this for me :D 

Comment 4 by carl...@google.com, Nov 7 2017

You're welcome!

But please note that there is the intention to remove this and other Navigation.TimeTo* histograms once PlzNavigate is fully landed. See issue 698957.

Sign in to add a comment