Use of Groups for Tabs Causes Freezup

I set up about a half dozen groups, problem seems unrelated to number of tabs per group. V1.35.103 (Feb 16, 2022)… Certain groups when accessed and tabs surfed cause the browser to lock up, have to kill it with ctrl alt delete… don’t recall any browser lockups before using groups

@Max2 ,

Requires study.

Using as tools that help you to see all of the processes involved – whichever of the following are available on your computing device:

  • OS Actvity Monitor
  • OS Task Manager
  • Brave Browser Task Manager
  • OS Performance Monitor
  • OS Resource Monitor

. . . and using for guidance:

Multi-Process Model Background

To understand Chromium’s memory usage, let’s understand the multi-process model. Unlike other browsers, Chromium is divided into multiple processes. When Chromium starts up, it will initially have two processes. One process is the browser process which controls the main browser functionality, and the other is the initial renderer process , which runs the Blink rendering engine and JavaScript (V8). Each time you open a new tab in Chromium, you’ll likely get a new renderer process. With typical browsing, it is common to see 5-7 chrome.exe processes active. Further, if you utilize plugins, apps, or extensions, they may also execute within independent processes. All of Chromium’s processes, whether it is a browser process, a renderer process, or a plugin process, will show under the Task Manager as “chrome.exe”.

Multi-process browsers were a mistake. They’re nothing but trash.

Is there anyway to launch Brave in single process mode without resorting to the command terminal, like setting a flag or something?

@Go-go_duck ,

"Brave" (Mac OS) appears in two locations:

/Applications/Brave Browser Framework.framework/Versions/

I have used the more senior of the two, when starting Brave Browser from the Mac OS window command line.

Brave.exe (Windows OS)

C:\Program Files\BraveSoftware\Brave-Browser\Application\brave.exe

Whichever you launch, those parent, or primary, processes (I do not know the correct terminology that the whiz-kids would use) are going to take steps that in turn, launch at least one of the member-processes of the aggregate.

Because, normally, to produce the initial web-type page window, there is a call(?) or some command(s) affecting the generation of a tab / window process . . . and that process shows up in the Activity Monitor / Task Manager.

In order to launch a primary process but not allow that primary to make calls . . .

If you have the skills and tools used by a developer, I imagine that they launch some of the individual processes (of the Brave aggregate), for testing, for example, at GitHub. IOW, the developers would have the ability to test just about any component of the whole.

Seems to me, in that case, you would need to use one or more of the command line flags among the Brave Browser list of hundreds of command line flags (date of publication, unknown):

Toggle the List of Command Line Flags/Switches

Your might discover something there, that gives you some ideas.

UPDATE (20220226 Saturday)

@Go-go_duck ,

Turns out, that in the list of command line flags, there is a Single Process flag, and I just tested it, by using the following command (on a Mac):

open -a “Brave” -n --args --single-process

Brave Browser did start up as a single process, but the initial window had the following warning:


PS. There are well over 1,000 of the command line flags.

1 Like

Thank you! I’ve been comparing Brave in multi-process and single-process mode over the past week, and the latter runs a lot smoother and especially faster on a 4 core, 8 thread CPU.
I’m using the --profiling-flush flag as well as the --single-process in case of crashes, though I’m not sure about the frequency with which profile data is saved by default. Anyone know?

Despite the warning about stability suffering, I’ve noticed no more crashes than otherwise (and significantly less time lost in restoring tabs), most of which seemed to go together opening a Private with Tor window, which, as you know, is already a whole ordeal in its own right.

What I’m more concerned about is the warning that security may, or rather will suffer. Do you have any idea what is meant by that?
I suspect they may just be saying that since they have not invested much effort into developing single-process mode. If I recall correctly off the top of my head, Firefox introduced tab containment before multi-process, though of course that does not mean Chromium (and by extension Brave) uses containers in single-process mode, merely that there is no reason that they couldn’t.

I really hope Brave starts to officially support single-process browsing, because for me it improves lag that all Chromium browsers suffer from, and which I’ve seen a lot of people complain about recently, by a lot.

@Go-go_duck ,

I am guessing, and “it seems to me,” that as much as possible, windows and tabs are intended to have a characteristic of being sandboxed, or as close to that as developers are presently able to affect.

Thus isolating windows from each other, to the extents of either possible or practical ( < that being a matter of possible vs convenience ). In other words, trying to limit cross-over connections to / from the Internet, between some windows (or tabs) but not others.

There are so many conditions that crop up, when connecting to a secure, financial website. You wish to sign on, to look at something confidential, but in the web page there happens to be a link to information – and you choose to read that info.

So, you click on the info link. Should Brave Browser produce a New Private Window, a New Window, a New Tab . . . all depending upon the level of security that you would wish, or expect, or Brave developers settled upon as a policy (for the moment, or long-term)?

If there is advertising at the Information web page, do you want to be certain that the advertisers are completely isolated from your previous (and still open) secure web page showing your financial details?

I do not know, if the practical mechanics involved, during --single-process allow for whatever kind of sandboxes (or sandbox-like) windows and / or tabs that you (and / or Brave developers) would expect, want.

I do assume, based upon Brave info, that there is better protection for each window / tab, when not using --single-process.

I do not know – but you might study – what happens to some Brave Browser > Settings when using --single-process.

You could use Developer Tools and watch the Network tool, to see what flows between website and a Brave Browser window, when using, or not use, --single-process.

NOTE, that a Developer Tools display, applies to the front-most Brave Browser window that existed when you called forth the Developer Tools. If you close that “Brave Browser window” then the “Developer Tools display” associated, will also close.

Also monitor the Developer Tools > Application – for cookie activity.

And, ask somebody at Brave, what they know.

Issues at GitHub, in which [issues] there is mention of “–single-process”"--single-process"&type=issues

An example issue:

Apropos, there, a screenshot:

Screen Shot 2022-03-03 at 6.49.46 PM

That (splits v spanning) may explain some of the background in Brave Browser behaviors, for which, I came up with workarounds; for example, re the order of opening new windows – in order to get a solid TOR window.

Update – Notice that the Brave extension exists for both

  • New Window process (“Extension: Brave”)
  • New Private Window process (“Private Extension: Brave”)

And when I took the following screenshot, no windows and no tabs were open . . . and, the “Private Tab . . .” refers to the last Private Window that was open . . . and the "Service Worker . . . " appeared in the list, long after no windows (and no tabs) were open:

Screen Shot 2022-03-03 at 10.15.37 PM

I have not found any reason why the Service Worker process appeared. I do not have any PWA (Progressive Web Applications) installed – as far as I know.

1 Like

Very interesting that the split extension processes for regular and private windows remain after closing the corresponding window. Especially if many extensions require use of the spanning model to function in at all (as per bridiver’s comment you screenshotted).

Like I said, I seem to recall Firefox supporting single-process sandboxing at some point, so I think it should be possible. But of course, that doesn’t mean that Chromium/Brave are structured in a way that allows it too, nor that it is as secure as multi-process sandboxing.

I haven’t gotten around to comparing network and cookie activity with dev tools in multi- and single-process modes yet. But I did notice just now that even the browser updates utilize multiple processes, which seems odd to me:

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.