This is actually fairly easy to reproduce. Open an icognito window to a page with a lot of links (Reddit, Hacker News, Google search results). Hold command and click on about ten links in a row (to open them in new tabs). 

Approximately one in every 3 or 4 will hang forever and never load the page. The stop and refresh buttons do not work when the tabs are stuck in that state.

Copying and pasting the url manually into a new tab will cause it to load instantly.

Deleted: Screenshot 2016-03-18 13.03.15.png 17.4 KB

	Screenshot 2016-03-18 13.03.15.png 17.4 KB View Download

Can you try creating a new user in chrome://settings and see if that user has the same problem (if not, you can just delete it). That will help see if any extensions are causing the issue.

If it does not solve the issue, can you please go to chrome://net-internals and try loading a page. After a minute of spinning, you can save the event log to a file and attach it here. Thanks!

Thanks for your reply. I tried creating a new user and the problem persists. It is intermittent, but I was able to capture a log from net-internals. It is attached.

It feels like some sort of race condition. Just for the record, I am using a computer that could probably be in a museum (MacBook Air 11 inch from October of 2010).

Also just wanted to point out that the issues with extensions failing to load was resolved when I rebooted my computer. 

Please let me know if you need any additional information.

Deleted: net-internals-log.json 255 KB

net-internals-log.json 255 KB View Download

I've had a similar (same) issue occur with the same setup.  My computer is also an older MacBook Pro (Early 2009/Core 2 Duo) "slower" system.  I had the problem with with OSX 10.9.5 suddenly.  I removed all Chrome files (including all Application Support and settings files), reinstalled, and still had the issue.

I upgraded to El Capitain (10.11.4) and the problem seems to have gone away with the exception that sometimes search from the url field doesn't do anything.

I noticed the NeedsFeedback label is still on this ticket. Can you please let me know what else you need from me. It is pretty frustrating cause I have not really been able to use Chrome reliably. I'd be happy to provide more info. 

If the plan is to drop support for OS X 10.9 Mavericks completely that is also fine, but I think you should have cut it off at Chrome 48 then. I believe Mavericks still is around 15% of Mac OS X traffic.

@iamcraigcam...: There's very little inside the net-internals; are you sure that you brought up about:net-internals, hit all the different links, waited until you were solidly in the hanging state, and then stopped collecting data and exported?  If so, can you give me some of the links you clicked on (preferably both ones that hung and ones that didn't), so that I can search for them in the net-internals dump?  Maybe while creating a new net-internals dump making sure you bracket the reproduction?

Your comment about the use of older computers makes me suspect some kind of memory pressure/crash; that's unfortunately more likely than I'd like.  But it's worthwhile trying to dig down and confirm that.

Thanks much in advance ...

rdsmith I definitely did it the same way you described. I am even able sometimes to get it to hang just by selecting Chrome -> About Google Chrome from the top nav. In the example net internals log I sent I only clicked a single link and was able to get the first one I clicked on to hang. 

I see the host in there when I open the file directly (did not see it when loading via the UI in chrome). The full URL I clicked on was:
http://yonomitt.com/blog/2016/3/23/meeting-people-is-easy-but-hard

Unfortunately, with this log I did not include a page that DID load successfully. I can take a new one tonight that contains a page that loads in addition to a page that hangs if that will be helpful.

Thanks for your reply.

@iamcraigcam...: When I bring up the .json file from c#4 in my favorite text editor (emacs, though I really hope that doesn't matter :-}) and search for "meeting-people" or "yono" (prefix for host) I don't find a match for those strings in the file.  Any chance you uploaded a different net-internals file than the one you're looking at?

It's definitely in there. If you click the view link and open it right in chrome and command f you will see it!

Deleted: 0JG4.png 232 KB

	0JG4.png 232 KB View Download

Right, sorry--not sure why my search didn't turn it up the first time.  What I find when I look at the net-internals dump for that host is what looks like pre-connects; the network stack is making connections to that host, but there is no URLRequest driving those connections or attached to them.  

When you get in this state, could you open up Chrome's task manager (triple line menu at the upper right of the browser window -> More Tools -> Task Manager) and see if you have processes corresponding to the tabs that you're launching and if they have reasonable numbers (some memory, some CPU, etc.)  If the network requests for this were being dispatched as preloads but the processes that are supposed to be created to send the actual requests aren't being created, that could explain it.  And that might fit with the "very old machine" aspect of what you're seeing.

@rdsmith The stuck tabs do, in fact, show up in the task manager. They show the URL and there is some memory usage, but I do not see any CPU usage.

After a very very very long time (maybe 5 minutes?) a notice comes up saying the page is unresponsive.

I have attached screenshots. Is there anything else I can do on my end to help?

Deleted: e2SZ.png 44.6 KB

	e2SZ.png 44.6 KB View Download

Deleted: kWBJ.png 26.5 KB

	kWBJ.png 26.5 KB View Download

I wanted to add that I just noticed that some of my extensions are failing to work properly. I mentioned that in the original description, but thought it was resolved. 

One in particular is JSON Formatter (https://chrome.google.com/webstore/detail/json-formatter/bcjindcccaagfpapjjmafapmmgkkhgoa?hl=en).

Both of the issues are present on the latest Chrome Canary build as well.

Would you be willing to disable those extensions and see if the problem persists?  (Or just see if you can reproduce it with a new, clean user; create from chrome://settings "Add person..."; that's an easy way to get a clean extensions slate.)

A hanging tab would make sense for causing this problem; I'm just not sure what would produce that, and it may be hard to track down :-J.

@rdsmith I already tried that if you look at #c4. I tried a clean user also in an incognito window (with all extensions disabled). The issue still happens. I think the loading issue also applies to extensions cause if I restart the browser the extensions will work some of the time.

Hi guys! Sorry if I'm butting in here, but does Guest mode happen to work for you? Your situation sounds very similar to the problem that we have with  issue 598088 , and if we use Guest mode, we don't have any problems with the tabs loading (might be a clue as to what's going on).

@aliao: No worries about butting in; a public bug tracker is for people to share ideas and solutions.  But "Browse as Guest" looks to me from looking at the code to be pretty much identical to incognito mode, which I'd think a new user would also mimic pretty closely.  And the OP has already tried Incognito mode; it was one of the early repro descriptions.  Does Incognito mode/New user *not* work as a workaround for your problem?

@iamcraig... : Sorry, missed that detail in c#4.  Ok, to summarize what we know so far: The URLRequest never gets created for links that are clicked on.  The processes associated with the request is created, and appears fine, but seems to hang.  I think we need to investigate why processes may hang, which is unfortunately the kind of general question that I'm not sure where to start with.  But I'll dig into it.

@rdsmith Unfortunately both incognito mode and creating a new user don't work for us, which makes it kind of peculiar that Guest mode works every time. It'd be interesting to see if Guest mode works for @iamcraig... :)

Hmmm.  Ok.  @iamcraig...., willing to try guest mode?  :-}

Yes. I don't have my 10.9.5 machine with me at work, but I will try it later tonight when I get home.

Wow. So yes. When I pick to browse as a guest everything works correctly, but incognito and creating a new user both fail as @aliao mentioned.

According to https://support.google.com/chrome/answer/6130773?hl=en the only difference between guest mode and incognito is that people browsing in guest mode “can't see or modify the computer owner's Chrome profile” whereas incognito can still access “history, bookmarks, passwords, Autofill data, and other Chrome settings”. 

I'm sure there are other differences under the hood, but maybe this is a place to start :\

Thank you for providing more feedback. Adding requester "rdsmith@chromium.org" for another review and adding "Needs-Review" label for tracking.

For more details visit https://sites.google.com/a/chromium.org/dev/issue-tracking/autotriage - Your friendly Sheriffbot

[Removing self as owner; triage duty is rotated automatically, and I'm off triage duty; the next net triager will pick it up.]

That is an interesting and weird thing.  Nasko, can you suggest how to route this bug? It seems to have to do with processes that are spun up for navigations hanging, and it happens in Incognito mode, but not in Guest mode.  

Seems like this is more of a navigation issue, like  issue 598088 .

I'm also on 10.9.5 and having the same problem. If it helps with the investigation, here's the net internals file after creating a new user and opening three tabs to www.google.com. Two of the tabs loaded and one stayed a spinner.

Deleted: net-internals-log.json 1.8 MB

net-internals-log.json 1.8 MB View Download

ysagal:  Thanks for the log!  That one shows 6 main frame requests for http://www.google.com, and 7 for the NTP's main resource.  Not sure if any of that's related to the bug, though.

A couple hanging requests, but none look relevant (4 are clearly hanging gets to get updates from servers.  Last is for docs offline, requested by an extension, I think - others may be, too).

One NTP request ends with SERVICE_WORKER_ERROR_KILLED, and others by SERVICE_WORKER_ERROR_KILLED_WITH_STREAM - but I assume that's all just you opening new tabs at the NTP, and then navigating away.

We see a bunch of requests to the same URL that have a number of SDCH errors - I suspect that's relevant, either, but seems concerning.

[rdsmith]:  Is it common to repeatedly use a cache entry that's returning a bunch of SDCH errors?  Seems like if that happens, we should evict the entry.  I suspect that SDCH is not the cause of the problem we're seeing, given the other logs of problems here indicate an issue spinning up the renderer, but all those SDCH errors are concerning.

Matt: That's WAI, and I don't think it's a problem; all those errors are just saying "We advertised an SDCH dictionary, but the server sent us unencoded content and didn't include X-SDCH-Encode: 0 to say they were doing it!!!  Oh, wait, that wasn't the server, that was the cache, from our first retrieval of that resources.  Never mind, everything's fine."

Do you understand why we're getting six main frame requests from opening three tabs?  I notice the service worker being involved in loading some of the tabs; is that relevant?  

Here's another net internals. I got a clean launch of chrome with one account, reproduced the bug by going to gmail.com and took a screenshot of the task manager. 3 tabs opened gmail.com just fine, one did not (pid 1027). Hopefully some commonalities will be helpful.

Deleted: net-internals-log (1).json 3.2 MB

net-internals-log (1).json 3.2 MB View Download

Deleted: image.png 126 KB

	image.png 126 KB View Download

The log shows on 3 gmail requests, so the 4th one never made it to the network stack.  It shows 4 NTP requests, so each new tab seems to have loaded, or at least tried to load, the NTP, just that one tab didn't load gmail.  There is a prerender for http://gmail.com/ that wasn't used.  Looking at the other log shows no prerenders for http://www.google.com, so odds are this is not a prerender issue.

Continue to think this is an issue starting up processes.  ysagal:  What CPU do you have?  There was a suggestion on another bug that this may only appear on Core 2 Duo CPUs.

Yes, I have a Core 2 Duo as well.

I also have this issue on my Core 2 Duo Mac Mini running OSX 10.9.5.

Going to go ahead and merge this with the other issue - it sounds to me like they're the same.  If not, we can split them up again.  Please watch that one if you're interested in updates.

Shouldn't the other one have been merged as a duplicate of this one ;)

I feel like people are more likely to find this issue since it is about pages intermittently failing to load which I think is the more common way this will manifest itself in the UI vs. pages not loading at all. Of course, I am biased since I made the ticket haha. 

As long as it gets fixed it is all good.

Also it was created a week before and has twice as many watchers

I merged this into the other one mostly because the loading team had commented there, and this issue looks to be more in their area than mine.  You're certainly right about the title of the other bug (Which I've fixed), the number of watchers, and the filing date.  It's just a matter of which issue I thought was a bit likely to get in front of the right people.

Cool, No worries! I just hope it gets resolved soon :)

I have the exact same issue, using 10.11.6 El Capitan on a MBP 15" Early 2011. I have had the problem for a couple months I think and was thinking Chrome update would get rid of it.
I uninstalled and reinstalled Chrome with no effect. 
Any solution to this issue yet?

I too am having a similar issue, new tabs just show the sad Chrome favicon and can't access any about pages as they do the same thing as well. I can't even tell you what version I have because of this stupid issue.

Mac, 10.11.6, Macbook Mid 2012