New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 899800 link

Starred by 1 user

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Android
Pri: 3
Type: Bug



Sign in to add a comment

Investigate higher loss rate of UMA data when users are offline

Project Member Reported by carlosk@chromium.org, Oct 29

Issue description

There have been evidence from collected UMA data that seem to indicate a bias of higher UMA data loss for users who are more frequently offline. UMA being one of the most important sources of data that drive decisions for Chrome development, this effect is an important issue to be investigated.

We want initially to confirm this issues does actually exist and attempt to measure how much it effects existing metrics. We'll limit the investigation to Android where this issue was initially identified.
 
Status: Assigned (was: Untriaged)
Cc: dtrainor@chromium.org
Hey Carlos, can you add some more context on how this issue was discovered?
Re #2:

* Initial report on this issue: go/collecting-uma-offline (view restricted)
* Public doc on Offline resilient metrics: https://goo.gl/GJXwFe
* Initial proposal of a resilient metric: go/offline-prefetch-metrics (view restricted)

And there's probably more information in email threads and the like.
Project Member

Comment 4 by bugdroid1@chromium.org, Oct 31

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/593985a892a32de785aad1688747363a0ff0e067

commit 593985a892a32de785aad1688747363a0ff0e067
Author: Carlos Knippschild <carlosk@chromium.org>
Date: Wed Oct 31 22:03:50 2018

Add OfflineUsage metric alternative that is not offline resilient

This change adds a new metrics that works as a mirror of
OfflinePages.OfflineUsage, aptly named
OfflinePages.OfflineUsage.NotOfflineResilient, that should report
exactly the same values but does not accumulate counters in
PrefsService, reporting them directly to UMA instead. We want to compare
these two metrics to try and assess the loss bias of metrics often
reported in offline situations.

Bug: 899800, 857134
Change-Id: I90b83ce1da638ca389ef38785a9f5ef1bb4538f3
Reviewed-on: https://chromium-review.googlesource.com/c/1305397
Commit-Queue: Carlos Knippschild <carlosk@chromium.org>
Reviewed-by: Ilya Sherman <isherman@chromium.org>
Reviewed-by: Dmitry Titov <dimich@chromium.org>
Cr-Commit-Position: refs/heads/master@{#604394}
[modify] https://crrev.com/593985a892a32de785aad1688747363a0ff0e067/chrome/browser/offline_pages/prefetch/offline_metrics_collector_impl.cc
[modify] https://crrev.com/593985a892a32de785aad1688747363a0ff0e067/chrome/browser/offline_pages/prefetch/offline_metrics_collector_impl.h
[modify] https://crrev.com/593985a892a32de785aad1688747363a0ff0e067/chrome/browser/offline_pages/prefetch/offline_metrics_collector_impl_unittest.cc
[modify] https://crrev.com/593985a892a32de785aad1688747363a0ff0e067/tools/metrics/histograms/histograms.xml

Sign in to add a comment