Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

grahamperrin

macrumors 601
Original poster
Jun 8, 2007
4,942
648
Spun off from Macbook Pro Late 2012 \ A1425 Model kernel_task

Background

I have, for test purposes, a MacBookPro8,2 – MacBook Pro (15-inch Early 2011).

MacBook Pro Repair Extension Program for Video Issues – in July 2015 the logic board was replaced by a reputable service provider, and since then I have rarely touched the Mac.

Also in July, it was realised that the battery had failed and was expanding, so it was removed. Whilst I test things, there's no rush to purchase a replacement battery.

Tests included installations of PC-BSD.

Sometimes neither the integral keyboard nor the integral trackpad is recognised (when this happens, USB keyboards and mice work fine). I wonder whether that's a side effect of attempting to boot, without a battery, after very long periods with no power (and the clock is sometimes wrong at startup after those periods). I wonder … but those symptoms don't bother me; they're not yet a focus area.

A week or so ago I performed Internet Recovery then, from Mac OS X Lion (10.7.x) an upgrade to OS X 10.11.1. Very little third party software – Mozilla Firefox, MacPorts and so on; nothing exotic.

On 2015-11-08 whilst using MacPorts to install GNOME Web (epiphany) and its dependencies, Activity Monitor often presented unrealistic measurements of the amounts of CPU used by kernel_task – 3,171.4 percent, for example.

If I recall correctly, installation of dependencies was interrupted by disturbance to the MagSafe connection.

The focus of this topic – the problem

At some point after I resumed installation of epiphany and its dependencies, Activity Monitor presented realistic measurements of inexplicably high use of the CPU by kernel_task. So high that it took more than three days to perform the installation.

The inexplicably high use:
  1. persisted long after use of MacPorts
  2. did not recur with a safe boot
  3. recurred after the safe boot.
I don't seek help with the problem – I don't plan to use El Capitan (or Yosemite) – but some of what I'll add to this topic might be useful to someone else who finds trouble with inexplicably high use of the CPU by kernel_task …

The possibility of a problem with hardware

On one occasion, with 10.11.1 neither OS X nor Recovery OS would start. During each start, an automated restart. Stuck in loops, of sorts. I did wonder whether both kernels were panicking, but the patterns of behaviour (including some verbose boots) made me suspect otherwise. When, eventually, OS X started normally I didn't seek a .panic file because I had reset NVRAM (more than once).

Of interest


CPU load constantly 100% MBP 13 inch Mid 2012 (2015-11-09), in particular:

… Power Bus (PBUS) … Any time the SMC cannot read a sensor it "protects" the hardware by running that kernel task, using up all your resources. There used to be a plist hack but that was on older pre-2012 models I believe. You either need to have someone fix your logic board (who knows what they are doing) or get it replaced by Apple. Apple won't repair the board... they only swap boards.

Usually sensor failures are related to liquid spills but not always.

No spills in my case.

I wonder whether the sense of absence of the battery is the trigger, but it's remarkable that the high use occurs only sometimes.

Of possible interest

The subject lines caught my eye, but I have not yet read the topics …
 
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.