Has El Capitan changed kernel panics?

A kernel panic. It is an interesting exercise to get your own screen shot of one. A kernel panic. It is an interesting exercise to get your own screen shot of one.

I have been hearing of quite a few Mac users who are running El Capitan, whose Macs periodically freeze, then spontaneously restart. My own iMac17,1 does this every few days, and has been doing so since I updated it to OS X 10.11.4. One thing which puzzles me about all these cases, including my own, is that no one seems to have experienced a classic kernel panic in El Capitan. By ‘classic’, I mean one in which a multilingual monochrome warning dialog is displayed before the restart.

It was only when I consulted Wikipedia’s excellent potted history of kernel panics in OS X, that I discovered that those ceased happening in OS X 10.8, which tells you how infrequently I have suffered them over the last four years. At least until being struck down by 10.11.4 in March 2016.

From OS X 10.2 to 10.7, when something catastrophic happened to the OS X kernel, a fault or failure of such severity that the only way to recover was to shut down or restart, one of the three different variants of the kernel panic dialog was displayed, and the user was then left to force the restart.

A kernel panic. It is an interesting exercise to get your own screen shot of one.

For the last four years, from OS X 10.8 onwards, Macs have behaved differently: they automagically restart themselves, displaying a kernel panic dialog when next starting up, for “a few seconds”. If your luck has really run out and there are five kernel panics within three minutes of the first, a prohibitory sign is shown on the display for thirty seconds, and the Mac shuts down and doesn’t attempt to restart again: that’s a ‘recurring kernel panic’, or more like a whole month of Friday 13ths.

Apple’s account of these issues is now quite different. They’re not kernel panics any more, but unexpected restarts, which makes them seem as innocuous as unexpected quits, perhaps. And you may not see any kernel panic dialog during the restart: “If your computer experiences a kernel panic, a message may appear for a few seconds explaining that the computer has been restarted”. Or that message may not appear, I suppose.

But according to Apple,
Once your Mac restarts successfully, an alert message appears, “You shut down your computer because of a problem.”
Oh no it doesn’t, not under OS X 10.11.4 and later. Then,
Once you log in, OS X lets you know that, “Your computer was restarted because of a problem.”
Which is something else that it does not do under OS X 10.11.4 and later.

Apple also informs us that kernel panic logs are located in /Library/Logs/DiagnosticReports. Although this iMac has been freezing and restarting every 2-5 days for the last five months, and there are 84 diagnostic reports in that folder, not one appears to relate to those kernel panics or ‘unexpected restarts’.

So I took this on to Apple’s Technical Note TN2063, Understanding and Debugging Kernel Panics. At least it is brave enough to call a spade a spade, and not fumble around with euphemisms. It’s a shame though that Apple doesn’t seem to have brought that Technical Note up to date since 2008 – yes, eight years ago – when kernel panics were still classic in nature.

So as far as I can tell now, El Capitan doesn’t undergo the classic kernel panic process, but just restarts, and may not give you any further information. When an ‘unexpected restart’ does occur, it may not leave any panic log, or send any such log or other information to Apple. So Apple may be blissfully unaware of all the grief that we are going through, all these Macs suffering ‘unexpected restarts’ every few days. The problems which should be swamping Apple’s automated support servers (which receive and process crash logs, etc.) may not even be on the support radar.

So here’s the important question: who has experienced a ‘proper’ kernel panic in El Capitan 10.11.4 or later? Did your Mac make you aware that it was sending the panic log to Apple? Or are we all victims of the most evil bug of all, which fails to create and send those panic logs?

I’d be very interested to know, please, so I can decide whether to start reporting all my ‘unexpected restarts’ myself.