New issue
Advanced search Search tips

Issue 734236 link

Starred by 1 user

Issue metadata

Status: Archived
Owner: ----
Closed: Aug 3
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug



Sign in to add a comment

Speed: memory leak on page in Chrome, not in Edge

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

Issue description

When browsing to this (somewhat unsafe for work) page memory grows unbounded. This page has crashed due to OOM twice when I've left it open for a long time.

http://www.thestranger.com/savage-love/2017/06/14/25209211/savage-love

A canary devtools heap snapshot is available (Google employees only) here:
https://drive.google.com/a/google.com/file/d/0B0_BQhEiKmSYdXh2ekdWbkx6am8/view?usp=sharing

A canary chrome://tracing memory-infra traces is attached that covers the time period when the process starts. If possible I will attach one from later one when memory has grown.

Note that the devtools heap snapshot was from a different run from the chrome://tracing trace, which is probably best given how much memory devtools can use.

 
trace_SaveLoveStartupTrace.json.gz
7.4 MB Download
Unfortunately the page is now behaving reasonably in canary and stable. I guess they fixed it, or else different ads are now being served. I'm not sure if the traces I got will show anything interesting.

It's quite possible that Edge was being served different content and that is why the leak didn't repro there. Should probably just close this. Sigh...
The numbers look pretty reasonable to me (see attachment).

Did the renderer crashed (OOM) in that trace?

How long did you ran the webpage before getting a memory snapshot?
metric.png
13.6 KB View Download
Cc: erikc...@chromium.org
I looked to every objects allocated more than 10k times.

Two processes contains some potential leaks (see attachement).

The CSS potential leak is tracked here:
  https://bugs.chromium.org/p/chromium/issues/detail?id=733714

The sync potential leak is trackec here:
  https://bugs.chromium.org/p/chromium/issues/detail?id=732969

sync.png
115 KB View Download
css.png
68.8 KB View Download
Unfortunately my traces were both shortly after refreshing the page. I'm still figuring out the best data to get.

It sounds like one of the best things to do is to get a devtools heap snapshot when the page is in its bloated state, whereas I was initially trying to use chrome://tracing with memory-infra.
Labels: -Performance Performance-Memory Needs-Investigation
Labels: Pri-2
Project Member

Comment 7 by sheriffbot@chromium.org, Aug 3

Status: Archived (was: Unconfirmed)
Issue has not been modified or commented on in the last 365 days, please re-open or file a new bug if this is still an issue.

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

Sign in to add a comment