MCA: CPU 1 COR (1) internal parity error

[no_connection]

After upgrading to FreeNAS-9.2.1.6-RELEASE-x64 (ddd1e39) I have started getting these errors.
I started using VB at the same time so it may or may not be related.
CPU usage does not seem to affect it as I ran OCCT in a VM with no error but one appeared at idle a while after.
I have not seen this error prior to upgrading. Previous was 9.2.1.5
I have not tried turning VB Jail off as I have running VMs but I will try that tomorrow.
Burn in prior to taking the server online did obviously not generate anything suspicious.
According to this thread it might be a bogus error:
http://lists.freebsd.org/pipermail/freebsd-questions/2014-January/255873.html

Code:

Jul  5 23:44:03 Chrysalis kernel: MCA: Bank 0, Status 0x90000040000f0005
Jul  5 23:44:03 Chrysalis kernel: MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000
Jul  5 23:44:03 Chrysalis kernel: MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 2
Jul  5 23:44:03 Chrysalis kernel: MCA: CPU 1 COR (1) internal parity error

 

[no_connection]

Dell T20 with 4+8GB ECC Dell UDIMM(Hynix)
4x4TB WD RED Z2
Extra Intel 100Mb PCI

Running extended diagnostics atm but nothing was logged in BIOS and quick test passed.

When done I will run the server without VB Jail started to see if errors continue.

Considering the OP in the link exchanged every part but the case and still got the same error as me points to a high possibility that it is not hardware malfunctioning.
Wonder if it is something related to Haswell architecture that might trigger the error.

*edit*
Test passed, I did a few more CPU stress tests and all passed.

 

[no_connection]

With no VB Jail started I have not gotten the error in 20 hours of uptime.

 

[dlavigne]

Anything in this thread help?

 

[no_connection]

I have tested the hardware and it seems to be fine, no further errors running without Jail. And as the other thread linked got the same error even with replaced hardware it might not be a real hardware error, even though the error itself says so. I need a bit more proof that it is a hardware error especially as all tests pass just fine.

I will turn on Jail again to see if errors start popping up.

 

[no_connection]

Started the VB Jail and let it run for a while with no VMs started and no error.
Within minutes of starting a VM the error appeared.

I made a USB stick with Intel diagnostics utility and will run that later.

Ether a CPU feature is broken and only manifests itself with very specific calls, although then it should probably return a different error.
Or something else related to VB or Jail is generating it.

Oh, the error itself is changing between CPU 0 and 1 and also give different APIC IDs

*edit* these are the errors this far today, first being minutes after starting a VM.

Code:

Jul  7 15:20:39 Chrysalis kernel: MCA: Bank 0, Status 0x90000040000f0005
Jul  7 15:20:39 Chrysalis kernel: MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000
Jul  7 15:20:39 Chrysalis kernel: MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 2
Jul  7 15:20:39 Chrysalis kernel: MCA: CPU 1 COR (1) internal parity error
Jul  7 15:20:41 Chrysalis kernel: MCA: Bank 0, Status 0x90000040000f0005
Jul  7 15:20:41 Chrysalis kernel: MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000
Jul  7 15:20:41 Chrysalis kernel: MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 2
Jul  7 15:20:41 Chrysalis kernel: MCA: CPU 1 COR (1) internal parity error
Jul  7 15:21:08 Chrysalis kernel: MCA: Bank 0, Status 0x90000040000f0005
Jul  7 15:21:08 Chrysalis kernel: MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000
Jul  7 15:21:08 Chrysalis kernel: MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 0
Jul  7 15:21:08 Chrysalis kernel: MCA: CPU 0 COR (1) internal parity error
Jul  7 15:41:13 Chrysalis kernel: MCA: Bank 0, Status 0x90000040000f0005
Jul  7 15:41:13 Chrysalis kernel: MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000
Jul  7 15:41:13 Chrysalis kernel: MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 2
Jul  7 15:41:13 Chrysalis kernel: MCA: CPU 1 COR (1) internal parity error
Jul  7 16:39:18 Chrysalis kernel: MCA: Bank 0, Status 0x90000040000f0005
Jul  7 16:39:18 Chrysalis kernel: MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000
Jul  7 16:39:18 Chrysalis kernel: MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 2
Jul  7 16:39:18 Chrysalis kernel: MCA: CPU 1 COR (1) internal parity error
Jul  7 16:39:18 Chrysalis kernel: MCA: Bank 0, Status 0x90000040000f0005
Jul  7 16:39:18 Chrysalis kernel: MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000
Jul  7 16:39:18 Chrysalis kernel: MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 2
Jul  7 16:39:18 Chrysalis kernel: MCA: CPU 1 COR (1) internal parity error

 

[no_connection]

20h since I shut down VM and no errors since.
Intel diagnostics all passed.

 

[no_connection]

An interesting observation, the error only appears to happen with a XP VM.
A very odd bug to say the least.

 

[Thor Miller Grotle]

I also have run into this problem running FreeNAS 9.2.1.6 on a new rig, meaning to run as a small fileserver with a RaidZ1 and a backup disk.

Running a jail with Virtualbox from preinstalled jail , everything seems to be running smoothly up until i start a virtual machine.

I have set a Windows 8.1, 32 bit VM. Running in bridged network /NAT mode.




When i deactivate the VM, I dont get the internal parity error

> pid 5082 (VBoxHeadless), uid 1001: exited on signal 10

> epair0b: promiscuous mode enabled

> MCA: Bank 0, Status 0x90000040000f0005

> MCA: Global Cap 0x0000000000000c09, Status 0x0000000000000000

> MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 6

> MCA: CPU 3 COR (1) internal parity error

> MCA: Bank 0, Status 0x90000040000f0005

> MCA: Global Cap 0x0000000000000c09, Status 0x0000000000000000

> MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 2

> MCA: CPU 1 COR (1) internal parity error

> MCA: Bank 0, Status 0x90000040000f0005

> MCA: Global Cap 0x0000000000000c09, Status 0x0000000000000000

> MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 4

> MCA: CPU 2 COR (1) internal parity error

> MCA: Bank 0, Status 0x90000040000f0005

> MCA: Global Cap 0x0000000000000c09, Status 0x0000000000000000

> MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 4

> MCA: CPU 2 COR (1) internal parity error

> MCA: Bank 0, Status 0x90000040000f0005

> MCA: Global Cap 0x0000000000000c09, Status 0x0000000000000000

> MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 4

> MCA: CPU 2 COR (1) internal parity error

> MCA: Bank 0, Status 0x90000040000f0005

> MCA: Global Cap 0x0000000000000c09, Status 0x0000000000000000

> MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 0

> MCA: CPU 0 COR (1) internal parity error

> epair0b: promiscuous mode disabled

> epair0b: promiscuous mode enabled

> MCA: Bank 0, Status 0x90000040000f0005

> MCA: Global Cap 0x0000000000000c09, Status 0x0000000000000000

> MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 2

> MCA: CPU 1 COR (1) internal parity error

Hardware:

MOBO: ASUS B85M-G, uATX

CPU: Intel® Core™ i5-4460

2x Kingston DDR3, 1600 MHz, CL11, 8 GB

6x 1TB WD10EADS

When running same VM on a HP Proliant Microserver gen7 (N54L) there seems to be no problem other than the lower CPU performance.

I plan to test a 64bit Windows 8.1 VM and a 64bit linux variant VM to test if it is a 32bit/64bit problem in Virtualbox.

 

[no_connection]

Running XP 32bit seems to generate the error almost immediately for me.
ESXi seems not to generate it, but nesting it under VB does not seem ideal, and I have problem getting XP to run without 100% CPU usage.
Ulteo also did not generate error.
I tried ZeroShell and it did generate one error.

So it seems 64bit VMs are free from the error while 32bit generates them.

The error shows up in main log so it might be worth taking a look at.
Ill submit a bug report, for reference if nothing else.
*edit* Bug #5607 created

 

[Thor Miller Grotle]

After upgraded my VM to a 64bit Windows 8.1, all CPU 0 COR (1) internal parity errors, have stopped.

 

[no_connection]

No way of knowing if it's a VB problem or Jail problem.
Ether way apparently we shouldn't be running VMs on FreeNAS in the first place so developers don't care.

 

[cyberjock]

Ether way apparently we shouldn't be running VMs on FreeNAS in the first place so developers don't care.

You know that the developers helped create the virtualbox features? So saying you "shouldn't" is kind of a slap in the face. And they definitely care because they *do* want it to work.

 

[no_connection]

Yeah. Um, don't do that. :)

And the "not to be fixed" kind of says the I shouldn't be doing it and they won't investigate further.
It was not a slap in their face, you said yourself that what happens in a Jail is not their problem.
"don't care" wad probably poor choice of words on my part, sorry about that.

 

[cyberjock]

Well, you are right that that they can't (or shouldn't depending on how you look at it) be responsible for fixing it. It's almost certainly a virtualbox bug and as we don't handle virtualbox stuff it's kind of out of our hands. ;)

If this bug were something that was our fault (the devs and I spent over a week getting that forsaken VM template to work) they'd be very interested in fixing it.

One thing you could try if you are determined to check things out is to try the Intel processor diagnostic tool... http://www.intel.com/support/processors/sb/CS-031726.htm

 

[no_connection]

Already done as I mentioned above, same with other testing tools, none reported any errors.

I did search quite a bit and the only relevant I could find was the thread I linked to where a user replaced *everything* and still had the error.
Thor above also has the same error doing the same thing, and I doubt he has faulty hardware ether.

If I had enough RAM on my ESXi box I would create a testbed and run FN with VB and see if that errors out.

I don't know how the error would look like of it came from inside the Jail, or what would get reported to main log (green one at the bottom).
But I don't thing anything virtualbox does should be reported by BSD as the error above.

If it was a VB thing alone it *should* be more wide spread, but I can't find much about it.

And at least two different users are now getting it while running 32bit VMs under VB from within FreeNAS.
So it could be specific to FreeNAS Jail in this case.

Or it's a Haswell thing that causes the error to pop up in this specific scenarios. It's a wildly used processor in my case G3220.
Have you run 32bit VMs without errors?

 

[cyberjock]

I've run Windows XP in a VM without problems. That was the OS I used when I was setting up the template everyone is using. I didn't do Windows 8 though since I hate Win8+.

Obviously the problem is very obscure. It could be some kind or problem with the FreeNAS jail + virtualbox + CPU + Guest OS. Unfortunately without a way to reproduce the error there's not likely to be a bug fix forthcoming.

 

[Rutgzven]

Got this as well.

Dell T20 with Pentium G3220 cpu.
FreeNAS-9.2.1.6-RELEASE-x64 (ddd1e39)

8GB RAM (Samsung m391b1g73qh0-yk0, Dell official memory for T20)
4GB RAM Hynix HMT451U7AFR8A-PB (included with server)

1x250 GB SATA drive, single drive pool, ZFS
1x500 GB SATA drive, single drive pool, ZFS
1x500 GB SATA drive, single drive pool, ZFS

System has been burnt in for almost two months.


A week ago I upgraded from 4 to 12 GB RAM, (run memtest86+ for a few hours after upgrade with no errors detected)
Some days later started using virtualbox jail as well.
Error started to appear after running 32 bit Windows server 2003 in virtualbox inside jail doing a lot of CPU/memory intense work.


dmesg printout:

Code:

...
GEOM_ELI:     Crypto: software                                                                                                   
vboxdrv: fAsync=0 offMin=0x37c offMax=0xddc                                                                                      
bridge0: Ethernet address: 02:85:55:f6:98:00                                                                                     
bridge0: link state changed to UP                                                                                                
em0: promiscuous mode enabled                                                                                                    
epair0a: Ethernet address: 02:62:a7:00:07:0a                                                                                     
epair0b: Ethernet address: 02:62:a7:00:08:0b                                                                                     
epair0a: link state changed to UP                                                                                                
epair0b: link state changed to UP                                                                                                
em0: link state changed to DOWN                                                                                                  
epair0a: promiscuous mode enabled                                                                                                
ng_ether_ifnet_arrival_event: can't re-name node epair0b                                                                         
em0: link state changed to UP                                                                                                    
epair0b: promiscuous mode enabled                                                                                                
epair0b: promiscuous mode disabled                                                                                               
epair0b: promiscuous mode enabled                                                                                                
pid 12447 (VBoxHeadless), uid 1001: exited on signal 11                                                                          
epair0b: promiscuous mode disabled                                                                                               
epair0b: promiscuous mode enabled                                                                                                
epair0b: promiscuous mode disabled                                                                                               
epair0b: promiscuous mode enabled                                                                                                
epair0b: promiscuous mode disabled                                                                                               
epair0b: promiscuous mode enabled                                                                                                
epair0b: promiscuous mode disabled                                                                                               
epair0b: promiscuous mode enabled                                                                                                
epair0b: promiscuous mode disabled                                                                                               
epair0b: promiscuous mode enabled                                                                                                
pid 13391 (VBoxHeadless), uid 1001: exited on signal 11                                                                          
epair0b: promiscuous mode disabled                                                                                               
epair0b: promiscuous mode enabled                                                                                                
pid 14248 (VBoxHeadless), uid 1001: exited on signal 11                                                                          
epair0b: promiscuous mode disabled                                                                                               
arp: 192.168.1.132 moved from 02:62:a7:00:07:0a to f8:b1:56:9b:c4:1e on epair0b                                                  
MCA: Bank 0, Status 0x90000040000f0005                                                                                           
MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000                                                                    
MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 0                                                                                
MCA: CPU 0 COR (1) internal parity error                                                                                         
MCA: Bank 0, Status 0x90000040000f0005                                                                                           
MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000                                                                    
MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 2                                                                                
MCA: CPU 1 COR (1) internal parity error                                                                                         
MCA: Bank 0, Status 0x90000040000f0005                                                                                           
MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000                                                                    
MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 0                                                                                
MCA: CPU 0 COR (1) internal parity error                                                                                         
MCA: Bank 0, Status 0x90000040000f0005                                                                                           
MCA: Global Cap 0x0000000000000c07, Status 0x0000000000000000                                                                    
MCA: Vendor "GenuineIntel", ID 0x306c3, APIC ID 0                                                                                
MCA: CPU 0 COR (1) internal parity error                                                                                         
[root@Dell_NAS ~]#                                  

I can also note that i have tried both installing/running the guest machine with one cpu and a guest machine with dual core cpu, and I can only get the Windows guest OS to run stable with single core virtual machine (and IO APIC disabled).

 

[andyclimb]

Right. I had this issue big time and posted a thread about it when running 9.2.1.6 beta. Since upgrading to the rc I've not had a single crash. I can't locate the thread at the moment as I'm on my phone. I was trying to run widows 7 and iTunes, and it would crash a lot. The mca cpu parity error also occurred.

Not got it since using the latest vb template too! So I'm happy. Although sorry it doesn't help you. But U are not alone that is for sure.

 

[Rutgzven]

New Dell T20 BIOS are available, released last week.

http://www.dell.com/support/home/us...e=poweredge-t20&languageCode=EN&categoryId=BI

Fixes & Enhancements
Fixes
- Fixed high CPU(Central Processor Unit) usage problem after the system is woken from the sleep mode.

Enhancements
- Update Intel Processor Microcode.
- Support Microsoft Windows Server 2008R2

I'm going to try this and see if this can help.