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

Issue 733894 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Aug 13
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 1
Type: Bug
Hotlist-MemoryInfra

Blocked on:
issue 690042
issue 733897
issue 733899



Sign in to add a comment

Investigate and reduce noise in system health memory metrics

Project Member Reported by perezju@chromium.org, Jun 16 2017

Issue description

Meta bug to keep track of efforts to analyze and reduce noise in memory infra metrics.
 
Cc: -nednguyen@chromium.org nedngu...@google.com
Cc: picksi@chromium.org
Blockedon: 733897
Blockedon: 733899
Cc: erikc...@chromium.org
Blockedon: 690042
Started doing a fresh analysis on noise [1], focusing on mobile, these are the metrics/stories with std error over 1MiB.

Across metrics, java_heap looks like one of the worst offenders, so will start looking there.

Across stories, the infinite_scroll appear to be the noisiest. But this is probably expected due to the difficulty of scrolling in a more deterministic way? Is there something we could do there?

[1]: https://colab.corp.google.com/v2/notebook#fileId=0B586tdnIdbhKbTNFM0hKSWhVTWs
metric_story.png
179 KB View Download
Components: Internals>Instrumentation>Memory
Labels: -Pri-1 Pri-3
As per our current OKR's reducing Pri on this issue.
Labels: -Pri-3 Pri-1
And re-raising again as this is now top Pri on my OKRS :)
Status: Fixed (was: Started)
I think this bug is not tracking much. :)

Sign in to add a comment