Kernel CPU use brings system to its knees....

Etresoft... thanks for your reading and advice; I'll address your comments.

EtreCheck does look at your system installation log. It's just that OS installs are special. The "install" is a download of the installer app.

Your EtreCheck report isn't listing kernel_task.

Those tasks are not listed as processes in Activity Monitor, so I figured it was EtreCheck splitting them out.

I will have to double-check that to see if it is still being detecting. I have to use some tricks to show kernel_task usage. But I still have some other tricks if that isn't showing up anymore. If you say the kernel_task is taking 800 % of the CPU, I'm not disputing that. That is typical behaviour of a system throttle.

Well, 800% is an extreme; it's usually only 500%. Do you see any other process using significant CPU or memory? I suspect an errant process in the Kernel; perhaps one of the system processes listed in the report. Even the trackpad and keyboard strain to get cycles; type a word, wait 3 seconds, then bang. Trivial tasks— selecting an app from the dock, forget it!

recommend a couple of things:

1) Never, ever run a "fan control" software. That kind of software will physically destroy your computer.

I understand you don't like it, but can you explain your assertions? What its doing to 'destroy' the computer? [ Etrecheck already decided it's trash. 😉]

I replied to Bob that it does not always (or, perhaps even 'usually') do this. It's intermittent, <not> ideal for diagnostics.

2) Disconnect your external display. The only time I've ever personally encountered this problem was when I had a bad dongle I had purchased on Amazon.

The monitor is connected with LG's supplied mini-Displayport to Display Port cable in a TB2 port on the Mac, and not series with anything. About 5-6 ft cable, no dongle. (I tried a cheapy cable just after getting the monitor (2 yrs ago), but it could not deliver a clean signal at any of the higher resolutions, at 60 Hz. I don't operate the monitor near its max capability; my eyes aren't that good!! 😆 (Shouldn't laugh.. recent retina surgery!)

3) Disconnect all other external hardware too, like disk drives.

4) Try a different charger

Yes, I could possibly do those things. How long does it take to determine that something intermittent like this is gone? I'd rather not buy another power supply (this one's still the OEM).

If the system is being throttled, then that could easily explain all the symptoms you are seeing. Your report took 56 minutes to run instead of 3 minutes. Your hard drive file system test failed. That normally takes 30 seconds but gets cancelled after 2 minutes.

I agree, the system is being throttled, by the kernel, not any user apps. When the kernel is behaving, the report takes - maybe 3 minutes. Etrecheck is running on the internal 512 GB APFS SSD, which tests fine- >1000MB/s R&W. Externals are 7200 rpm 4 TB Toshibas, APFS via TB2 (10Gb/s). Diagnostic sw shows none have any HW errors, re-allocated blocks, Read or Write errors, etc.

The ultimate cause of this problem is almost always a hardware failure. If you're lucky, it is some external hardware that you can just unplug for an instant fix.

Can a 'failure' fix itself? For security, the external drives are attached only when running TM or CCC, or saving an archive. Can't say one way or the other whether it's happened then, but pretty sure it has.

If you're not lucky, it is internal hardware like the logic board.

See above.

Or maybe the battery? 9 cycles on a 2015? That compute is just barely not on Apple's recently updated list of vintage and obsolete computers, so Apple should still repair it if you are willing to invest the money.

I changed the batteries as a matter of safety. System Information showed full capacity, and no complaints, but the computer was rocking on the case bottom. All 4 batteries in the set were badly swollen. Thus, the 9 cycles. The replacements are (OWC) aftermarket because a search did not located any OEs. I avoid e-Bay...

Many thnaks for your thoughts. I'd especially like to understand what's throttling the computer.

Allen_31, thanks for your suggestions; I also hoped they might help., but I believe I've run those gauntlets. Closely compare the two attached EtreCheck runs, both with the same monitors, resolutions, external drives, start-up items, and normal apps. [Safari and Reunion, a Mac-only app.] Safari has many suspended tabs as usual. The first report was run when kernel cpu use was 100's+ %; the second when normal - a few % to occasionally to a few tens of %.

The idea of hardware failure was based on the first run, when EtreCheck couldn't get enough cpu cycles.. The second report, when kernel use in normal, shows User CPU load to be low compared to top system items during kernel runaway.

I read that CPU temperatures in the low-mid 60s C is normal, and I do use Macs Fan Control to keep the system cooler than normal. Fans start at lower temperature, and has lower maximums. A cooler system can only improve reliability, other than possibly wearing out the fans — which are trivial to replace on this machine. Thermocouple temperature readings at the fan exhausts should show fan failure; under normal conditions yesterday, exhaust temperature was is in the 100 - 110 °F range, both left and right. Fans and radiator were carefully cleaned ~ 10 months ago, which isn't a problem since our cats have died.

However, I am exploring an interesting claim that the system interprets maximum fan rpm to mean more cooling is required, and that it then runs "heatless" kernel cpu cycles to reduce cpu availability to User, to reduce temperature. If true, then aggressive cooling setpoints could conceivably contribute. Curious logic - reduce fan speeds to lower temperature.

Other confirmable items you suggested have been done - SMC and NVRAM reset; RT ~ 78°F., low battery charge (runs on AC); two Apple Hardware tests, — no problem. Restarting has only temporarily tamed the kernel when it was high... but how does one really know? The machine's been running normally for several days, but I doubt it's 'fixed'. 😉

Cheers! Keep 'em coming.

Thanks, Bob, for your kind reply.

The logic board had been professionally replaced about 4 years ago, but the battery was not new. I recently had to replace it, and carefully cleaned fans and heat exchangers with paint brush and filtered air.

I haven't touched any IC thermal coupling, or paste, even though my background tempted me. I was not aware the system could throttle so badly, which poor chip coupling could do. But that would not be intermittent.

It certainly does not do this all the time. Been running the same computer, monitors, ext'l TM drive, and same apps —and same fan control— as yesterday, since 7 a.m. today and never saw the kernel above ~10% — and that just temporarily.

It's been running about 3-4% today. But once in a while, it gets into a funk.

Thanks for your thoughts, Bob.

I cleaned fans and the radiator with a small paint brush and air duster when I replaced the swollen batteries.

EtreCheck clearly shows the most intense processes are Apple's, save for Safari. A valid hypothesis has to deal what factors might cause it to be intermittent. I haven't seen one yet.

A technical group I worked with always cited 'sunspots'. 😀

Yesterday and toda, about 20 hours in total, the computer ran all normal applications, external disks, and typical memory loads without a hitch. The attached report shows how the computer usually behaves except for the 7-day look-back impairment with periods of the kernel issue. Feel free to contrast it with the report first posted.

So here I pause, I doubt very much that it's hardware. I suspect there's some aspect of software - who knows what? — this computer serves cache and Handoff with 2 other computers and 2 iPhones, sharing 200 GB iCloud among them.

No more gaunlet! 👍😄

I'll give that a try, John... next time it acts up. Meanwhile, will check whther anything I'd want to run won't.