Nonstop Parity Errors

[BinaryPatrick]

I've posted about this before and thought I had it solved. Every time I run a parity check, I have parity errors. It seems like the longer I go between checks the more errors I have, but nonetheless I have at least 6K - 90K errors every time. At this point I have checked the cabling, run Memtest86 about 4 different times (both in parallel and serial across all CPU cores), and run SMART tests on the drives. No errors are detected in any of my testing. I feel like I must have a bad drive. When I set up this server 2 months ago, I was a n00b and didn't preclear at all.

 

Should I break apart my config, start over, and preclear all the drives / run extended SMART tests? Are there any other tests I could run at this point? I have my data backed up on other devices so I purge my unraid instance at this point if I need to. 

 

I'm running 3x 2TB drives running xfs.

[JorgeB]

Post diags after at least two parity checks with errors (without rebooting).

[BinaryPatrick]

22 hours ago, johnnie.black said:

Post diags after at least two parity checks with errors (without rebooting).

Yea I don't think I can get two without errors. There are always errors, that's my point. 

 

homeserve-diagnostics-20200520-1308.zip

[JorgeB]

The syslog posted is filled with crashes, but I can't see the beginning of the problem or parity checks, reboot, start a parity, let it finish, do another one, then post diags, without rebooting.

[BinaryPatrick]

Do you know what those crashes are? My googling didn't give any hints other than some Kernel problem..

[JorgeB]

13 hours ago, johnnie.black said:

can't see the beginning of the problem

Might be easier to see what's causing them if the syslog covers the start of the issue.

[BinaryPatrick]

I ran extended SMART tests on all HDDs and it returned no errors, but running another parity check resulted in 846,120 errors. Attached are the diagnostics. Syslog was 100% full.

homeserve-diagnostics-20200522-0244.zip

[itimpi]

Your syslog is full of errors that start with

May 22 00:14:03 homeserve kernel: DMAR: ERROR: DMA PTE for vPFN 0xf2827 already set (to 200000000000 not 758ed9002)
May 22 00:14:03 homeserve kernel: ------------[ cut here ]------------
May 22 00:14:03 homeserve kernel: WARNING: CPU: 2 PID: 21150 at drivers/iommu/intel-iommu.c:2300 __domain_mapping+0x205/0x2dd

Not sure what causes this, but they should not be happening on a well-behaved system.

[JorgeB]

Looks like a hardware problem, run memtest and/or try a different board/controller if available.

[Trucking Geek]

I am having this same issue with a new system.  (i7-12700K, MSI Z690i AUROS Ultra, 32GB of Kingston Fury DDR5, 3x 1TB Samsung 980 Pro NVMe.)

 

System seems to run fine for a few hours, then log fills up with the 

kernel: DMAR: ERROR: DMA PTE for vPFN 0xf2827 already set (to 200000000000 not 758ed9002)

errors.

 

Did you ever find a resolution?  I am letting it run again now after lowering the memory clocks.