[turtle]

My Mac mini keeps rebooting itself at odd and random times. I actually installed Zabbix on it so I would be really depressed to know how many times it reboots itself. I doesn't have a hard workload nor has it ever really had more than "office" tasks. The hardest it has now is being my primary Plex server.



It is the 2.5GHz i5 model with RAM upgraded to 16GB. Overnight tonight the machine rebooted itself 8 times. Normally I would expect to see the "Report issue" dialogue or something but I never get it. The machine just chimes the Mac boot chime and loads up the user profile that's set to auto-login.

Where can I look to track down the issue? I know I can take it to an Apple Store to be looked at but I would rather do what I can from home first. Surely there are some logs in there somewhere that will give me a clue.

I have done a clean OS install on it with a erase and install of it. The version is even up to date (10.12.3) though the version of the OS doesn't effect the rebooting at all. I even changed the name of the machine to Vanellope in honor of the "glitch" from Wreck-It Ralph".

[turtle]

Putting those two graphs together makes it look like the CPU overheats or something under heavy load. I don't see a load of 2 on that CPU as a heavy load, but that is interesting.

[alcimedes]

Yeah, I'd crack the case and clean out the guts, likely dust is causing the cooling to no longer be sufficient.

Then I'd check to see if you have some unintended background process going on that's causing those spikes to begin with.

[kscherer]

Mac Mini's vent from the air right on the surface of your desk, which is also where all the dust bunnies live. Al is likely right. Probably dust accumulation on the heatsinks causing overheating.

You know what to do!

[turtle]

What is that command that forced a CPU to 100%? It is like do something > /dev/null

There is no reason a short spike in load should result in a reboot. I will be taking this apart later this weekend though that is for sure.

[Dorian Gray]

I’d go so far as to suspect a broken fan, though enough dust would have a similar effect.

[alcimedes]

Originally Posted by turtle What is that command that forced a CPU to 100%? It is like do something > /dev/null There is no reason a short spike in load should result in a reboot. I will be taking this apart later this weekend though that is for sure.

yes > /dev/null &

once for each processor.

$ killall yes

when done.

[PB PM]

Originally Posted by Dorian Gray I’d go so far as to suspect a broken fan, though enough dust would have a similar effect.

That is possible, the Mini only has one wimpy little fan, and if it goes you'd be in trouble real quick.

If it's not a bad fan, try using a third party fan controller software which will ramp up the fans sooner to prevent overheating. I use Mac's Fan Control on all my Mac's and works like a charm to keep things cool, since Apple's fan curve is far to conservative.

[alcimedes]

in my experience though, unless you heard noises from the fan prior to overheating, the fan is still working fine. clean the sucker out really well, i'd guess it will be just fine.

[turtle]

So it's not a thermal issue at all. I'm running each thread at 100% and have been for over 30 minutes and no reboot. I heard the fan ramp up as expected and it sounds like a jet now, but it still hasn't shut down. Something else is causing the restarts.

So what about a log, surely this kind of thing gets logged somewhere?

Edit: Temp monitoring stuff. I found a link to a tool from Intel to monitor the temps and was able to verify I have it at the thermal limit.

[turtle]

I let the CPU run at 100% for almost 29 hours and there was not one single shutdown/restart! This is somewhat frustrating.


Is there a memory diagnostic I can run on this? Like memtest86+? What kind of hardware tests do I have access to without going to the GB? At this point I'm thinking I'll take it there and see if they can figure it out. I just don't want them telling me to do a clean install of the OS again.

[Dorian Gray]

Glad the fan is spinning at least. Did you do these tests with the unit in the same location as it was when suffering shut-downs?

[Eugene]

Doesn't matter if your CPU is at 100% just running "yes." It's not really performing any hard tasks. Download something like Prime95 and run a blend test. Also 95C is quite hot for a modern Intel CPU...past the point where it would start throttling itself.

[turtle]

I haven't moved the computer at all before running yes. I'll try Prime95 on it. It would be nice to know exactly what's going on with it.

Still hasn't rebooted yet either.

[alcimedes]

The next step I'd take then would be to dive into the Console.app and start looking at the logs prior to the shutdowns. Odds are there will be some recurring issue you'll see there. That process can take a while though, and you'll find some recurring themes that look suspicious but turn out to be nothing.

So the other odd thing, are you saying that when the machine was being completely thrashed for the test, it didn't reboot for 24+ hours in a row? That's very unusual given what you were seeing before, isn't it?

Maybe something related to waking from sleep? (I've seen where the sleepimage gets corrupted and it craps the bed when waking up, kill the sleepimage and no more issues).

The fact it was running something non-stop changed whatever state was causing the reboots. That to me is an interesting clue.

[turtle]

The machine shouldn't be sleeping since it's a Plex server. However it is possible the disk is able to go to sleep. There is also a USB disk attached that it does TM backups to. That might be where this is happening from.

I've completed two hours of Prime95 and there is no difference either:


Edit: I'm running memtest86+ on the machine now. I'll likely just leave it going for the next day or so too. Assuming this passes then I'll look into TM and disks sleeping.

[alcimedes]

Is it connected directly to your TV via an HDMI cable? This person's problem sounds somewhat similar, but they tracked it down to turning their connected TV off and on.

https://discussions.apple.com/thread...art=0&tstart=0

The last comment on the page mentions it happens to them when the machine wakes from sleep. I'd be willing to bet if you set your machine to never sleep, and never spin down the drives, the problem goes away. (that or turn off the require password on screensaver)

[turtle]

I do have the machine set to require a password for the screensaver. That could fit too. Not sure what the setting is for it, as in how long until it turns on.

The mini is connected to an old computer via HDMI -> DVI so that should be it.

[PB PM]

Originally Posted by Eugene Doesn't matter if your CPU is at 100% just running "yes." It's not really performing any hard tasks. Download something like Prime95 and run a blend test. Also 95C is quite hot for a modern Intel CPU...past the point where it would start throttling itself.

Indeed. My 2012 Min never gets that hot, I think the highest I've seen is 60ºC, even under long loads rendering video files. 97ºC (first picture) is getting very close to not just throttling (throttling starts around 85ºC IIRC), but thermal shutdown for that class of lower TDP dual Core i5 chips. I'd be tempted to tare it down and check to see if the heatsink is mounted properly.

[turtle]

The memtest86 completed without issue. I'm not surprised at all by this. So I was looking up the CPU since you all mentioned it and this particular CPU has a max temp of 100DegC. So clearly this thing is running hot. Idol it is running at almost 70DegC. I haven't found others talking about their CPU temps with this machine.

Maybe there is a member of this forum who works with people who work on Apple machines that might know what is considered normal for a MacMini5,2 with i5 2.5GHz CPU in it...

I'm still contemplating opening this up and replacing the thermal paste.

For now, I've removed the requirement for password on the screensaver and disabled putting the drives to sleep. Up next is look at the TM backup logs and see what shows if anything. I'll see how these two changes work in keeping the machine from rebooting.

[Eugene]

Intel has digital sensors for both core temperatures and package/heatspreader temperature. That 100C value is for the cores, and I bet that single number recorded by Power Gadget is the heatspreader.

Heatspreader tempsmshould be 5-10C lower. Either way your mini is running close to or over the max rated temperature.

[turtle]

Originally Posted by Eugene Intel has digital sensors for both core temperatures and package/heatspreader temperature. That 100C value is for the cores, and I bet that single number recorded by Power Gadget is the heatspreader. Heatspreader tempsmshould be 5-10C lower. Either way your mini is running close to or over the max rated temperature.

Yeah, it runs hot for sure. I was looking around trying to find some others talking about their temps and only found one article on Mac Rumors about it. Seems upper 90's isn't that uncommon for my model. I'm still thinking I need to see about reseating the fan/thermal paste. My room temp is high though, the office I'm in sadly sits around 80DegF/26DegC.

[PB PM]

26ºC isn't unreasonable, even for work. In the summer my workspace is often 30-32ºC, and never see temps like that. That said the 2012 model I have is the newer 2.5Ghz Dual Core i5, (Ivy Bridge, 3210M), so it's not directly comparable sadly.

My iMac has a Sandy Bridge chip, like your Mac Mini, and never runs hot like that, but it's a quad core and the iMac has better cooling.

[turtle]

After my earlier testing and the two changes I made the mini rebooted itself twice this morning. Going through the logs now. Currently there is no password on the screensaver, though it was still enabled. Now I think I'll have no screen saver and just power off the monitor.

The HDD were also set to not sleep, but I'll be looking through logs to see if I can find the cause.

[turtle]

So far I've only figured out that spike in CPU is the follow up from the reboot. All the Apps (like Plex) starting up. It happens after the reboot, not just prior to the reboot.

Quicklook is the only thing I've seen in the logs just before the reboots today:

Feb  8 05:35:08 Vanellope com.apple.xpc.launchd[1] (com.apple.quicklook[28117]): Endpoint has been activated through legacy launch(3) APIs. Please switch to XPC or bootstrap_check_in(): com.apple.quicklookFeb  8 05:39:19 Vanellope com.apple.xpc.launchd[1] (com.apple.quicklook[28174]): Endpoint has been activated through legacy launch(3) APIs. Please switch to XPC or bootstrap_check_in(): com.apple.quicklookFeb  8 05:39:19 Vanellope syslogd[38]: ASL Sender StatisticsFeb  8 05:40:40 localhost bootlog[0]: BOOT_TIME 1486550440 0

Feb  8 07:04:59 Vanellope com.apple.xpc.launchd[1] (com.apple.quicklook[1615]): Endpoint has been activated through legacy launch(3) APIs. Please switch to XPC or bootstrap_check_in(): com.apple.quicklookFeb  8 07:07:46 Vanellope com.apple.xpc.launchd[1] (com.apple.imfoundation.IMRemoteURLConnectionAgent): Unknown key for integer: _DirtyJetsamMemoryLimitFeb  8 07:07:47 Vanellope logd[54]: _handle_cache_delete_with_urgency(0x7f8746d3efd0, 3, 0)Feb  8 07:07:47 Vanellope com.apple.xpc.launchd[1] (com.apple.imfoundation.IMRemoteURLConnectionAgent): Unknown key for integer: _DirtyJetsamMemoryLimitFeb  8 07:07:47 Vanellope com.apple.xpc.launchd[1] (com.apple.quicklook[1680]): Endpoint has been activated through legacy launch(3) APIs. Please switch to XPC or bootstrap_check_in(): com.apple.quicklookFeb  8 07:07:48 Vanellope com.apple.xpc.launchd[1] (com.apple.TMHelperAgent.SetupOffer): Service only ran for 0 seconds. Pushing respawn out by 10 seconds.Feb  8 07:07:48 Vanellope com.apple.xpc.launchd[1] (com.apple.quicklook): Service only ran for 0 seconds. Pushing respawn out by 1 seconds.Feb  8 07:07:49 Vanellope com.apple.xpc.launchd[1] (com.apple.quicklook[1689]): Endpoint has been activated through legacy launch(3) APIs. Please switch to XPC or bootstrap_check_in(): com.apple.quicklookFeb  8 07:07:51 Vanellope com.apple.xpc.launchd[1] (com.apple.mdworker.shared.02000000-0000-0000-0000-000000000000): Service only ran for 7 seconds. Pushing respawn out by 3 seconds.Feb  8 07:10:22 localhost bootlog[0]: BOOT_TIME 1486555822 0

[turtle]

She has rebooted herself 17 times today. No reason that I can find either. None at all. I'm just really annoyed about this. I guess I can remove the USB drive and see if it has any impact on this though it's the Time Machine drive.

Screensaver is off completely. Drives are not set to sleep when possible.

/me frustrated

[Eugene]

Was the mini more stable when you were running synthetic loads on it? This might be a SpeedStep voltage switching/drop issue.

[turtle]

Yeah, it was way more stable when running the CPU at 100% with "yes". Today I've done nothing with it at all and yet it's rebooted itself 20 times up to this point. Zabbix data, console logs and any other data I can find doesn't give a clue of what's causing the reboot.

If it is the voltage switching then what? New Mac? Hardware repair at the Apple store out of warranty?

[PB PM]

It could be one of two things if there are voltage issues, either the CPU itself is going the way of the Dodo, or it could be a sign that the power supply is going. Now the real trouble is that the latter could have caused the former.

Do you have the machine attached to a UPS by any chance? Some cheaper UPS's without pure or simulated sin waves can cause problems with modern (post 2006) computer power supplies.

If it's just the power supply it's most likely wouldn't be outlandish to replace it. IIRC the 2011 model has an integrated PSU like the newer models, so if that is the problem you can try to get the part and install it yourself. If it's the CPU, you'd need a new logicboard, which wouldn't be worth while since it would cost almost as much as a new model (from Apple I suspect).

[turtle]

I wonder if the GB will be able to run a diagnostic on it that would tell it those kinds of things. I guess I'll finally schedule an appointment with it and see what they can tell me. I hate to lose this machine, but it certainly isn't very useful right now. For the last three days the average uptime is just over one hour.


I scheduled my appointment for next Saturday to see what they say. Hopefully it isn't just that the machine is out of warranty. I already know that. I'm hoping they will be able to run diagnostics on it that find something, anything. At this point I wouldn't mind a definitive "it's dead/dying" so I can be don't with it.