I had this problem on my Mid 2009 MacBook Pro / 15" / 8 GB, but I'm also a PC engineer.. This seemed & looked a lot like graphic / GPU problems.
My MacBook Pro has an NVDIA 9400 series graphic processor, I went to NVidia's website, and they have a CUDA driver download for OSX Mountain Lion. I did the download, installed it, and I now have an applet in Sys Preferences.
I also have MUCH greater video & graphic performance, no stickies & freezes when typing, etc. I had the spinning beach balls all the time, etc. All is now gone, no flickers, no freezing (so far), etc.
Console log looks different & better now as well.
I recommend doing the following:
- Get rid of Chrome & Firefox temporarily
- Clear cookies
- Download CUDA driver
- Disable Sys Pref / security settings to allow unknown developer software.
- Install CUDA
- Fix Permissions
I have almost never used Safari, so I have this happen even on cases where Safari's never been opened since boot.
I have a pretty reliable reproducer (use external display, run Minecraft, log into game, toggle fullscreen). I am thinking of doing a fresh install of 10.8.2 on new media and checking whether it happens there; if it doesn't, then I have at least some sort of bounds on the issue.
Oh, I'm in no way asserting that only video games can trigger this, just that it can be triggered when Safari isn't running.
And I have an update: I have found convincing evidence that a clean install would likely fix this. I have a Mini that's running 10.8.2. Same build as the MBP. So I put it in target disk mode, booted to its disk, and ran my tests.
No problems at all.
1. A clean install of 10.8 is likely to fix it.
2. If your machine came with 10.7, and you recover to that, then install the update, it may not fix it.
Well, normally I try not to reply to my own posts, but:
1. Merely re-running the "install OS X" from the recovery partition did NOT fix it.
2. Reformatting and reinstalling did.
However, this isn't quite a complete test, because I haven't re-loaded all my stuff yet. Migration Assistant is on that now, and thinks it has about 2 hours left. If it still works after reloading, then at least one cause of this (there may be more than one) is a bug in the 10.8.2 upgrade install over 10.7. I've saved a copy of the old System folder, and there shall be SUCH comparisons made.
Interesting. I am really curious as to whether anyone can find a specific change that triggers this.
It's been absolutely 100% reproducible on my existing install, since 5 minutes after I set the system up. I'm trying to keep notes updated on my bug filed with Apple.
At this point, I think it is totally unambiguous that something is broken in 10.8.2 or its upgrade, but I have no idea how to find out what specifically.
10.8.3 is out to developers... The focus area for their input is Airport, Safari, and Graphic drivers. Putting in the NVidia GPU sub-processing drivers (which enable the video card GPU to do it's own math calculations), seemed to make the problem much more subtle, but didn't go away entirely, it still froze after "sleep" and other innocuous things.
Interestingly, I have a friend at work with a 13" Macbook Pro / same era as mine (Mid-2010) along with my son's MacBook Air of about the same lineage as well, and neither have the problem, but both are using SSD drives and I have a Western Digital Scorpio Black / 750 GB.
You're not breaking the kernel... based on what Apple has developers focussing on right now for compatibility... I suspect the OS is getting pre-occupied with some subsystem, probably TCP/IP protocol stack or maybe IP version 6, and it goes away into never-never land. Usually, if you wait long enough, it comes back, but that's obviously unusable.
Microsoft converted rather quietly, but relatively successfully, to using IPV6 as the core protocol for Windows Server 2008, and if people removed it from the card binding, it would get pretty weird pretty quickly in certain areas - particularly if using Microsoft Server Clusters, the heartbeat channel for the server heads to monitor each other relies on IPV6 and if only using IPV4, it would kind of burp & sputter and act weirdly. Even if you didn't use IP V6, it needed to be on there and doing nothing, basically, or subsystems would start to hang randomly (kind of sounds familiar). They had to do it earlier-on for the Windows Server IIS support for websites. Same with Linux/Apache server. Later service packs though addressed the hanging issues and it generally tolerates stupid-configurations.
I'm wondering if Apple isn't having to do the same thing for future compatibility and it's causing some issues on some devices with specific network card chipsets or something, which would be a different unique driver for each. Graphics even touches that as the current generation of projectors for conference rooms are usually IP-based, rather than using the stupid old VGA cables up the wall or anything like that, so on a mobile laptop graphic card, there would be IP (and now IPV6) support built in somewhere.
It's one thing to support the new protocol with a simple binding for it or whatever, but it's another to rebuild the core OS to use it as a primary..
No idea, either way, after messing with it for about a week, I decided I have more important things to worry about than whether or not my MacBook Pro is working, so I back-revved for now.
I kind of doubt my ISP supports IPV6 yet though, and if it is centered on something like that, I would be in the camp with problems. It seemed like it generally worked a little better when experimenting and sitting in a coffee shop, but I wouldn't make the claim or guarantee it, I was too transfixed on figuring out the issue.
What the heck are you talking about?
1. This is video drivers, not networking stack. The hang I'm talking about is the one where the GPU crashes and the driver doesn't recover well. You can see tons of NVDA errors in system.log, describing this.
2. No matter what, if a user app can kill the system completely like that, there is a kernel bug.
3. If the kernel is fine without my files restored, and crashy with them restored, then something in the files that were restored has broken the kernel.
"Graphics even touches that as the current generation of projectors for conference rooms are usually IP-based, rather than using the stupid old VGA cables up the wall or anything like that, so on a mobile laptop graphic card, there would be IP (and now IPV6) support built in somewhere."
... No. This is not how computers work. Please stop talking.
EDIT: Now that's awesome. I had made a second account to run Migration Assistant from. I got curious. Tried on that account. No hang.
So something that is happening in my account's login process is triggering this...
Well, this is disturbing:
I removed the existing Application Support directory for minecraft, re-ran it so it re-downloaded all its stuff, and it worked. This is disturbing, because I had previously reviewed all that stuff, freshly downloaded it, and done many other experiments with it. I'd replaced things with newer things, replaced them with the default downloads, and so on.
And no matter what it's doing, a user app shouldn't be able to kill the OS.
That's a function of the process scheduler... it can and is tuned differently between workstation versions and server versions of the OS product. You don't have a single-user ancient Mac OS... what you have is BSD Unix with a Mac OS interface on it.
I don't know anything about computers though, and I'm busy putting up the 24 inch marble tile I bought with some of my end-of-project bonus on my paid vacation after finishing the $110 million dollar enterprise application development project I just finished and I have to pick up the AR-15 later today that I bought with some of it as well, and the $1800 Miele dishwasher I bought my wife a couple of days ago.
Go back to playing with your Minecraft game. Have fun. When it wears you out, start looking at how multiuser unix systems work, and how inter-related the operating system is with IP protocol, and how something as simple as a broken DNS client can hang the system pretty severely.