New issue
Advanced search Search tips

Issue 904884 link

Starred by 2 users

Issue metadata

Status: Duplicate
Merged: issue 879220
Owner: ----
Closed: Nov 16
EstimatedDays: ----
NextAction: ----
OS: Mac
Pri: 2
Type: Bug-Regression



Sign in to add a comment

Chrome randomly stops being able to load content, and cannot be killed without full system reboot

Reported by ja...@queso.com, Nov 13

Issue description

UserAgent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_13_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/70.0.3538.102 Safari/537.36

Steps to reproduce the problem:
1. Use Chrome normally

What is the expected behavior?
Chrome will continue to function, loading content as requested

What went wrong?
For the past two or so weeks, my installation of Chrome has seemingly-randomly gone into a state where it won't load content into any new tabs or windows at all. The entire set of symptoms is weird and confusing to me, so I'll just list what I see:

1. Chrome the app itself largely still remains responsive — I can create a new tab, I can interact with Chrome's menus, and I can even reload content in many (but not *all*) existing tabs. For example, I almost always have Gmail open in a tab, and Gmail will faithfully let me refresh my inbox, or switch to a different folder/label.

2. If I open a new tab or window and type a URL in, the browser just sits and spins; Chrome's own little loading spinner spins clockwise without ever loading any content, and will do so forever (I've let it go for an hour or more).

3. While I can interact with Chrome itself, I cannot open up the dev tools — meaning I can't tell if my page request in a new tab is actually sent out on the wire at all, or if the issue is prior to Chrome even doing that.

4. If I quit Chrome (using CMD-Q), the open windows will disappear, but the Chrome app itself will not quit; the icon remains live in the Dock, and the app itself remains running.

5. If I force-quit Chrome, a bunch of Chrome-related processes remain running. I can do a "ps -ax" at the command line, and there usually will remain about 6-12 Google Chrome Helper processes, as well as the crashpad_handler process. Not infrequently, there also remain a few "Google Chrome" processes as well.

6. I can "kill -9 (pid)" the crashpad_handler process and it dies. I can also almost always do the same for any "Google Chrome" processes that remain running. But at times for these, and almost 100% of the time for the Google Chrome Helper processes, if I "kill -9" them, then they become zombies — they remain in the process list with parentheses around them, and are completely unkillable. If I dump a process tree (pstree), they are owned by launchd, so I can't even kill their parent.

8. Restarting Chrome itself doesn't help; the new instance of Chrome behaves just as the force-quit one did, with no tabs/windows able to load any content. The only way to return Chrome to usable is to fully reboot the Mac. (Of note, I can't even successfully shut DOWN the Mac when this happens. Going through the normal shutdown process just leads to an interminable hang as the OS is trying to cleanly shut down; I end up having to force-power-off the machine and then power it back on.)

9. Safari and Firefox continue to operate normally when Chrome goes into this state — e.g., it isn't an issue that arises from some inability to use the network.

10. This has happened to me when I have one or two windows with 25-30 tabs total in them; it's also happened to me when I only have a single window and three tabs open. It doesn't appear to be related to specific content, but it's impossible for me to figure that part out.

I'm not sure what else I can describe to help, but this is clearly causing issues for me! I just went through and culled a bunch of extensions, not that I have any specific clue that an extension was at fault. The extensions that now remain are:

* 1Password
* AJAX Debugger
* Authy Chrome Extension
* Boomerang for Gmail
* Chromoji
* Cisco Webex Content Sharing
* Cisco Webex Extension
* Form Filler
* GitHub Custom Tab Size
* Google Docs Offline
* Hasher
* IPvFoo
* jQuery Audit
* JSON Viewer
* Mercury Reader
* Node.js V8 --inspector Manager
* Protect My Choices
* Tabs Backup & Restore
* Video Downloader professional
* Web Developer
* XV -- XML Viewer

and then the Chrome apps that I have installed are:

* Authy
* Docs
* Postman
* Sheets
* Slides

Can I help with any other information?

Crashed report ID: 

How much crashed? Whole browser

Is it a problem with a plugin? N/A 

Did this work before? Yes unclear

Chrome version: 70.0.3538.102  Channel: stable
OS Version: OS X 10.13.6
Flash Version:
 
A few questions:

- Does your computer have Symantec DLP installed?
- Does the New Tab Page load, or is it also blank when you open a new tab?
- When Chrome is in this state, could you grab a CPU sample for the renderer process that is hung? http://dev.chromium.org/developers/how-tos/debugging-on-os-x#TOC-Taking-CPU-Samples
Holy *crap* -- I was about to say "nope, no Symantec DLP" but then I decided to do my due diligence and track down how I'd even KNOW I have Symantec DLP installed. And you know what? The laptop *does* have it installed! Which is embarrassing; I had seen  issue #879220  when searching for bugs that could be related to what I am seeing, and had dismissed it because I honestly had no clue our enterprise had deployed Symantec DLP.

For your second question, I'll pay attention next time — but I'm 99% sure that the New Tab page does NOT load when I open a new tab; it's just blank.

And yep, I'll get a CPU sample next time it happens.
Yeah, if it's Symantec DLP then  issue 879220  is the root cause. Apparently upgrading the version should fix it. A CPU sample from a hung renderer would confirm that issue.
I think that at this point this issue can be considered a dupe of  issue #879220 , so much that that issue convinced our enterprise to disable Symantec DLP until its effect on Mac Chrome can be better-understood and remediated.
Mergedinto: 879220
Status: Duplicate (was: Unconfirmed)
Apparently there's an update to DLP that does not have the issue, but I don't have any version information unfortunately.

Sign in to add a comment