New issue
Advanced search Search tips

Issue 855724 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Jun 2018
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 2
Type: Bug-Regression



Sign in to add a comment

Component Updater Success Percentage Regression

Project Member Reported by waff...@chromium.org, Jun 22 2018

Issue description

Investigate success percentage regression shown by

https://uma.googleplex.com/p/chrome/timeline_v2/?sid=b37de8580597bed5d2b48fbce65ba0cf

Seems like a server-side or Finch-based trigger.
 
This is due to an increased number of RETRY_LATERS, triggered by a server-side release on May 17th or 18th.

Before: We would issue Retry-After (10 hours) to any request that doesn't have an update-check from 0.0.0.0.

After: We issue a Retry-After (10 hours) to any request that is background priority.

I suspect the M67 rollout is when we added the explicit "bg" header. In M66, we would have fallen back to our fg/bg heuristics, which includes the update-check-from-0.0.0.0 as sufficient to promote the traffic to fg, which is why for M66 the old is equivalent to the new.

I believe we are issuing this header by mistake anyways, and will drop it from the config within the next hour (after I confirm expected QPS impact).
Status: Fixed (was: Assigned)

Sign in to add a comment