macpro keeps shutting down

Question

Level 1

43 points

macpro keeps shutting down

My MacPro 3,1 has begun to act very wonky. I can’t pinpoint it to any specific event (new hardware, new software, new preference config) so I need some help deducing where the errors are originating.

Symptoms:

Turn the computer on, it can run for a minute, it can run for 15, but eventually it will just without warning shut down. Sometimes it will reboot to the login screen. Other times it will chime 2 or 3 times and get ‘stuck’ somewhere leaving the power light on the power button blinking, which requires a hard shut down (hold the power button until it turns off). It doesn’t seem to matter what task I’m doing to trigger these symptoms; I have been surfing the web, filing in Finder, live streaming on OBS. Not one consistent repeatable action.

Methods:

So far I have tried, repeatedly in some cases, the following to troubleshoot this problem. Reset SMC by disconnecting the power cable for 15 minutes. Also tried holding the power button down after resetting SMC to get the loud beep at first boot (not sure what this task is called). Zapped the PRAM by holding CMD-OPT-P-R at boot until the second chime. Unplugged external USB devices connected to a USB3 PCI card. None of these improved the symptoms. I have 3 user accounts on this Mac and this behavior is present when booted into all 3 accounts. Ran First Aid in Disk Utility on the boot drive while booted into Recovery Mode. Ran fsck in Terminal at the same time.

I did boot into Safe Mode by holding down Shift at boot. This helped. I was able to use the computer in limited function without any problems. Obviously I was not able to do any significant work as certain computer functions would not perform correctly (video display was very chuggy due to the NVIDIA driver not being loaded, audio would not play due to my sound interface being plugged into the USB3 card). So this suggests to me it’s a software issue, a kernel extension maybe?

Conditions:

I am running El Capitan 10.11.6. I have an NVIDIA GTX760 PCI card and I keep the OE ATI card for booting into Recovery Mode situations, but is disconnected in typical daily use. I have a Sabrent 4-port USB3 PCI card, that connects video capture, audio interface and SSD drives. The 3 built-in USB ports are connected to a Macally keyboard (that thinking about it now, is the only new item that I added right as this started happening), a USB hub on the Dell monitor, and a Sabrent 5-port USB2 hub. I have 32GB RAM in an 8x4 configuration.

I have a Console crash report showing the kernel panic of the most recent crash. I have not captured any previous reports to compare if the same trigger was the cause. I will paste it’s con below.

I think I covered everything but can provide any further details on request. Please if you have any ideas to troubleshoot this I’m in desperate need. My productivity has slowed to a crawl. I doubt the so-called Genius at the Apple store can help with low level deep in the weeds stuff like this, they’re more suited to helping Mum and dad get their iTunes set up, from my previous experience.

thank

adam

Mac Pro, OS X 10.11

Posted on May 3, 2021 11:04 AM

Reply

Answer 1

Top-ranking reply

Adam Mercado Author

Level 1

43 points

May 3, 2021 11:11 AM in response to Adam Mercado

Panic Report ***
Machine-check capabilities: 0x0000000000000806
 family: 6 model: 23 stepping: 6 microcode: 1551
 signature: 0x10676
 Intel(R) Xeon(R) CPU           E5462  @ 2.80GHz
 6 error-reporting banks
Processor 0: IA32_MCG_STATUS: 0x0000000000000004
 IA32_MC0_STATUS(0x401): 0xf200000410000800
 IA32_MC5_STATUS(0x415): 0xf200001044100e0f
Processor 1: IA32_MCG_STATUS: 0x0000000000000004
 IA32_MC0_STATUS(0x401): 0xf200000410000800
 IA32_MC5_STATUS(0x415): 0xf200001040100e0f
Processor 2: IA32_MCG_STATUS: 0x0000000000000005
 IA32_MC5_STATUS(0x415): 0xb200000084200e0f
Processor 3: IA32_MCG_STATUS: 0x0000000000000005
 IA32_MC5_STATUS(0x415): 0xb200000080200e0f
panic(cpu 1 caller 0xffffff8009fd075a): "Machine Check at 0xffffff7f8c44b103, registers:\n" "CR0: 0x000000008001003b, CR2: 0x0000000109e77000, CR3: 0x000000000d535000, CR4: 0x00000000000026e0\n" "RAX: 0x0000000000000000, RBX: 0xffffff80288a9000, RCX: 0x0000000000000001, RDX: 0xffffff7f8c468e20\n" "RSP: 0xffffff91f3eabd40, RBP: 0xffffff91f3eabd70, RSI: 0x0000000000000001, RDI: 0xffffff7f8c45fb90\n" "R8:  0xffffff80288ac400, R9:  0xffffff7f8c460110, R10: 0x000000c9775fe2fa, R11: 0x00000000e0000000\n" "R12: 0xffffff80281a0800, R13: 0x0000000000000000, R14: 0x0000000000000148, R15: 0x0000000000000001\n" "RFL: 0x0000000000010046, RIP: 0xffffff7f8c44b103, CS:  0x0000000000000008, SS:  0x0000000000000010\n" "Error code: 0x0000000000000000\n"@/Library/Caches/com.apple.xbs/Sources/xnu/xnu-3248.73.11/osfmk/i386/trap_native.c:168
Backtrace (CPU 1), Frame : Return Address
0xffffff8009d4acf0 : 0xffffff8009edab52 
0xffffff8009d4ad70 : 0xffffff8009fd075a 
0xffffff8009d4aed0 : 0xffffff8009fec48f 
0xffffff91f3eabd70 : 0xffffff7f8c44433a 
0xffffff91f3eabe60 : 0xffffff7f8c443690 
0xffffff91f3eabf20 : 0xffffff8009fd1a29 
0xffffff91f3eabf40 : 0xffffff8009ef22fd 
0xffffff91f3eabf90 : 0xffffff8009ef27f0 
0xffffff91f3eabfb0 : 0xffffff8009fca3f7 
      Kernel Extensions in backtrace:
         com.apple.driver.AppleIntelCPUPowerManagement(218.0)[917C14A9-67C6-3D62-B175-F802514D060C]@0xffffff7f8c441000->0xffffff7f8c46bfff

BSD process name corresponding to current thread: kernel_task

Mac OS version:
15G22010

Kernel version:
Darwin Kernel Version 15.6.0: Thu Jun 21 20:07:40 PDT 2018; root:xnu-3248.73.11~1/RELEASE_X86_64
Kernel UUID: 7564B0E7-EB5D-3887-BA79-59C870165AB1
Kernel slide:     0x0000000009c00000
Kernel text base: 0xffffff8009e00000
__HIB  text base: 0xffffff8009d00000
System model name: MacPro3,1 (Mac-F42C88C8)

System uptime in nanoseconds: 865791336665
last loaded kext at 246555121258: com.coriolis-systems.driver.Snapshot	123 (addr 0xffffff7f8cc38000, size 585728)
last unloaded kext at 307019269472: com.apple.filesystems.msdosfs	1.10 (addr 0xffffff7f8cc29000, size 61440)

Reply

Answer 2

Adam Mercado Author

Level 1

43 points

May 4, 2021 2:00 PM in response to Grant Bennet-Alder

Okay, thanks Grant. Good info. Agreed, fix one problem at a time. Process of elimination.

So, further to the ongoing situation. This is really weird. Pulled Riser A completely out of the Mac. Only Riser B installed. Boots fine. Runs fine. Cannot get it to crash after running Photoshop, QuickTime and others. The 4 DIMM LEDs light up at boot then go out. I noticed 1 red LED on Riser A so I guess that will indicate the bad slot.

Now, I thought the 3,1 needed RAM sticks in matching pairs across both Risers. All RAM present and accounted for in system profiler. So the 3,1 works more like 4,1 after all?

Also FWIW all LEDs on the DIAG button showing good

Reply

Answer 3

May 3, 2021 11:22 AM in response to Adam Mercado

Kernel panic, machine check, detected by multiple processors, error code 4 or 5 -- correlates very strongly with...

ECC memory errors.

User Tip: Mac Pro and Error Correcting Co… - Apple Community

.

Reply

Answer 4

Adam Mercado Author

Level 1

43 points

May 3, 2021 5:57 PM in response to Adam Mercado

Thank you Grant. That’s very interesting. I bought my first Mac in 1996 and it never ceases to amaze me how many more fascinating technical tidbits i can learn about.

Incidentaly the Mac ran smoothly all day, after the boot crash boot crash boot crash tango first thing. Looking at the system profile I see that 16GB of the 32GB installed is missing. The entire riser B is showing as empty. Could all 4 sticks be bad or something with the riser itself. The chances of all the sticks on one side going bad seems suspicious. But thanks for pointing out where to look and confirming the problem. I guess the RAM almost passed muster on boot up and once running just took a dump, until it couldn’t pass the ECC boot test at which point it’s not used, and the system works perfectly.

This RAM is not too old either, only a couple of years, from a reputable maker too. OWC is my usual go to. Sucks.

I will pull pull things out and do some tests tonight.

Reply

Answer 5

Adam Mercado Author

Level 1

43 points

May 5, 2021 10:18 AM in response to Adam Mercado

Quick update. Spent yesterday testing the RAM, two at a time. Turns out it’s all fine after all. Go figure. After reseating all the modules and the risers no DIMM could produce an error. All 32GB shows up in system profiler. No LEDs lit on the riser. Quite strange.

Just getting everything cleaned out and reseated must have must have done the trick.

Still baffled by the fact that I could run the Mac with only one riser installed. I’m certain 3,1 models need RAM matched across both risers. If anyone has any insight into this I’d love to hear it.

Also, Grant did you mention that there is some place to find the number of error bits the ECC has captured. So we can view the DIMM going bad. I don’t see anything in system profiler. I’m on El Cap so maybe this was introduced in later versions?

thanks

adam

Reply

Answer 6

May 3, 2021 6:32 PM in response to Adam Mercado

On that era Mac Pro, when you populate all slots on a riser-card, the memory pairs combine in a way that makes that memory riser-card slightly faster.

As I said in my little article, the ECC Hardware is applied very aggressively at Startup, and ANY errors that occur during the very brief self-test cause the entire module to be declared not present. Its partner may be fine, but BOTH will be declared not present.

If the pairs are doubled up, that may mean the entire lot of them are declared not present. I suggest you try testing each pair separately, if you can.

Although there is some fast logic on the riser-card, the likelihood of the riser-card failing is dwarfed by the likelihood of a module failing.

Reply

Answer 7

May 4, 2021 6:36 AM in response to Adam Mercado

Your kernel panic report is unambiguous. You have a RAM memory problem. But as usual, it does not tell you exactly which DIMM(s) are the problem.

It is conceivable that one DIMM could make a pair not appear, and having that pair not appear would knock their partner pair on a full card off the air as well.

Reply

Answer 8

Adam Mercado Author

Level 1

43 points

May 4, 2021 9:24 AM in response to Grant Bennet-Alder

I pulled my MacPro out of the office to test finally. Typically it’s right after I just reconfigured, and rerouted all the cabling so it was all nice and tidy.

Anyway im systematically going through and testing pairs of DIMMs

I still don’t understand how booting into safe mode will unequivocally and without fail or fault keep the machine running without error. What kernel extensions cause the crash, or unloaded don’t cause the crash? Is the ECC memory check part of some kernel extension or framework that just gets skipped? But wouldn’t bad RAM just be bad RAM kext or no kext? I’m not disputing your assertion that I have a bad stick somewhere, the system profiler confirms that. I’m struggling to understand the safe boot role in the process

cheers

adam

Reply

Answer 9

May 4, 2021 9:48 AM in response to Adam Mercado

If your DIMMs have deteriorated slightly from the condition you reported initially, it is possible that safe mode is benefitting from having slots declared "empty", early in the boot-up process.

Whether memory slots are declared "Empty" due to initial errors is determined by the ROM-resident Power-On Self Test.

My attitude is, "Fix the problems I know about, because the symptom may change afterward (and any other problem may then come into sharp focus)."

Reply

Answer 10

Adam Mercado Author

Level 1

43 points

May 4, 2021 12:06 AM in response to Grant Bennet-Alder

Yes that’s my understanding the DIMMs are installed in pairs. So it’s possible 2 sticks are bad and their respective partners are ignored at boot up too? I did not get chance to pull things out and test tonight. But I did battle a number repetitive crash reboots. Which again was cured by booting into safe mode.

So, how does safe mode ignore ECC, bad memory, or other faulty hardware? Dos that not suggest a software issue?

Reply

Answer 11

May 5, 2021 10:58 AM in response to Adam Mercado

<< Quick update. Spent yesterday testing the RAM, two at a time. Turns out it’s all fine after all. >>

Au contraire, you have what diagnosticians call, "No Fault Found", which is not the same as "All is Well."

--------

The summary you seek is NOT on the About this Mac > Memory ... display.

 Menu > About this Mac > (System Report) > Memory

User Tip: Mac Pro and Error Correcting Co… - Apple Community

That display is static. To get new data, choose Refresh or invoke the display again.

.

Reply

macpro keeps shutting down

Similar questions