September 25Sep 25 I've been meaning to post a forum question for well maybe over a year.UnRAID 7.2.4 - installed on a Intel NUC 11 Extreme Kit (NUC11BTMi9) Mini PC - Core i9-11900KB, using Samsung 4TB NVMe, for cache.It was stable, and then I started getting complete restarts, when using VMs, and I suspected it was because I was using nested ESXi.Earlier I decided to upgrade to 7.3.2, and to my amazement, managed to get 29 days of uptime with no issues or restarts, which was a bonus, as previously it would restart at least every two days, I then decided it was about time to setup a new syslog to harvest all the messages, so I could refer to later.But, recently I noticed the all the VMs had stopped, but oddly the server had not restarted.I noticed BTRFS errors on the cache, and it was put into read only mode, and hence all the VMs stopped.BTRFS error (device nvme2n1p1 state EA): parent transid verify failed on logical 4693510684672 mirror 2 wanted 585721 found 585705I copied all the data off the NVMe devices, and found no errors, and on restart and the scrub found no errors.Checking syslog, I can see this sort of activity recently2026-09-24T03:02:50+01:00 root: /mnt/cache: 212.4 GiB (228075569152 bytes) trimmed on /dev/nvme2n1p12026-09-24T03:02:50+01:00 kernel: I/O error, dev loop2, sector 85536 op 0x3:(DISCARD) flags 0x800 phys_seg 1 prio class 22026-09-24T03:02:50+01:00 kernel: BTRFS warning (device loop2 state EA): failed to trim 1 block group(s), last error -52026-09-24T03:02:50+01:00 kernel: I/O error, dev loop2, sector 85536 op 0x3:(DISCARD) flags 0x800 phys_seg 1 prio class 22026-09-24T03:02:50+01:00 kernel: BTRFS warning (device loop2 state EA): failed to trim 1 block group(s), last error -52026-09-24T03:02:50+01:00 kernel: I/O error, dev loop3, sector 37832 op 0x3:(DISCARD) flags 0x800 phys_seg 1 prio class 22026-09-24T03:02:50+01:00 kernel: I/O error, dev loop3, sector 75776 op 0x3:(DISCARD) flags 0x800 phys_seg 1 prio class 22026-09-24T03:02:50+01:00 kernel: BTRFS warning (device loop3 state EA): failed to trim 2 block group(s), last error -5I also notice MCE errors, but I've ran memtest86+ for over 24 hours, and no issues are detected with memory.It would seem since, restarting today 25th, after memtest86+ all is well, and after the scrub found no errors, just trying to identify why it went offline.btrfs scrub status /mnt/cacheUUID: e859fbd1-6099-4943-8c4f-3cb5f630dca8Scrub started: Fri Sep 25 11:18:10 2026Status: finishedDuration: 0:22:15Total to scrub: 5.97TiBRate: 4.58GiB/sError summary: no errors foundbecause at present its not been a very stable server, and I'm trying to consolidate 7 Synology NAS, and two Storage Servers, and many ESXi servers and home labs - to reduce electricity costs. But don't have issues with Synology, Storage Servers and ESXi servers in terms of stability. It could be a faulty Intel NUC 11 Extreme Kit (NUC11BTMi9) Mini PC or not suitable, but would rather not waste my time, if this is going to be a long term issue, don't get me wrong had UnRAID running on other mobos with no issues for years.Comments ?
October 20Oct 20 Author So again, failed with similar issueTitle: btrfs RAID1 cache repeatedly forced read-only (“parent transid verify failed” on both mirrors), Unraid 7.3.2SystemUnraid 7.3.2, upgraded from 7.2.4 on 25 Aug. 7.2.4 had occasional unexplained reboots.Intel NUC 11 Extreme (NUC11BTMi9, i9-11900KB), BIOS DBTGL579.0073 (Nov 2024)2×32GB Kingston DDR4-3200 SODIMM (KF432S20IBK2/64), non-ECC, running at rated speedCache: btrfs RAID1 on 2× Samsung 990 PRO 4TB, firmware 8B2QJXD7. Holds VMs (domains), appdata, docker.img and libvirt.img.What happenedAbout 16 Sep: cache forced read-only. parent transid verify failed ... wanted 585721 found 585705 on mirror 1 and mirror 2, state EA. I copied the data off, then rebooted, and the pool mounted read-write. I didn’t reformat it. It then ran normally.19 Oct, about 16:04 UTC: the same failure after 24 days of uptime, with no reboot, crash or power loss beforehand. wanted 654838 found 654812 on both mirrors. Running VMs then logged Read-only file system I/O errors every second.cyrusnas5-diagnostics-20261020-1300-redacted.zipSyslog server has now been setup, but its hosted by UnRaid, so has gone offline, when cache went read only!I'm now wondering if Cached NVMe, is really causing an issue, and despite the mirror if one fails, is causing more issues than a single. Edited October 20Oct 20 by einsteinagogo
October 20Oct 20 The logs show corruption before the later parent transid errors. September’s log includes metadata corruption detected while preparing a write, and October’s invalid-address errors started on October 17, before the pool went read-only.Both cache SSDs currently report zero media errors and zero error-log entries. That does not exclude a storage fault, but there is no clear evidence that one failed SSD is bringing down the mirror. Switching to a single SSD would reduce redundancy without addressing the cause.The machine-check notifications deserve investigation alongside the write-time corruption. RAM, the CPU or its memory controller could be involved, although these logs do not identify a failed component. The earlier clean Memtest86+ result does not rule out an intermittent fault.Keep Docker and the VMs stopped while you secure a current, verified backup on separate storage. Do you have any decoded machine-check records from your monitoring tools? The diagnostics contain the notifications, but not the details needed to identify the affected subsystem.Also confirm whether XMP, CPU overclocking or undervolting is enabled. Further memory testing should use stock settings, testing each DIMM separately. Put the syslog receiver on a system that does not depend on this cache.A writable mount after reboot and a clean scrub do not establish that the original problem was fixed. I would avoid btrfs check --repair or reformatting at this stage.
October 20Oct 20 Author ThanksThere are 20 MCE notifications in total: 16 between 2 and 16 Sep (around the first failure), none during the stable weeks, then 4 between 19 Oct 21:25 and 20 Oct 09:01.Decoded record from 20 Oct 09:01: Corrected error, MCA: Internal parity error, STATUS 9000004000050005, CPU 7, BANK 0, Intel family 6 model 141 step 1, microcode 0x58. The September records were lost at a reboot;BIOS: [no XMP, overclocking or undervolting / or what you find]. The RAM runs at its rated 3200 MT/s.Memtest86+ has already been run for 48 hours with each DIMM tested separately, with no errors. This was after the first incident.Decoded MCE records from /var/log/mcelog (attached): five corrected Internal parity error events, all in bank 0. Four are on logical CPU 7 (11 Oct 18:21; 20 Oct 00:15, 00:35 and 09:01) and one on CPU 3 (19 Oct 21:25). CPU: i9-11900KB, family 6 model 141 step 1, microcode 0x58. Given repeated bank 0 errors on the same core,I'm wondering if the CPU is hitting thermal limits, and this maybe is the issue with a certain core. Time to re-think its location. Edited October 20Oct 20 by einsteinagogo
October 20Oct 20 Author I decode the logs, and provided the info.but here there are cat /boot/logs/mcelog-20261020.txtmcelog: failed to prefill DIMM database from DMI dataKernel does not support page offline interfaceHardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC 1da3408449827TIME 1790488970 Sun Sep 27 07:02:50 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC 44b04fb2db871TIME 1790697001 Tue Sep 29 16:50:01 2026MCG status:MCi status:Error overflowCorrected errorError enabledMCA: Internal parity errorSTATUS d000008000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC 5ac63ede669f5TIME 1790814655 Thu Oct 1 01:30:55 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC 8aab35381b0d1TIME 1791069796 Sun Oct 4 00:23:16 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC b1b61a20b1175TIME 1791277782 Tue Oct 6 10:09:42 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC c0d2b8f23f905TIME 1791358285 Wed Oct 7 08:31:25 2026MCG status:MCi status:Error overflowCorrected errorError enabledMCA: Internal parity errorSTATUS d000010000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC db7a5e2537bebTIME 1791500279 Thu Oct 8 23:57:59 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC fc012ab83b3b1TIME 1791673554 Sun Oct 11 00:05:54 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC 108584d2e17795TIME 1791739293 Sun Oct 11 18:21:33 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 3 BANK 0 TSC 18c2977cb761edTIME 1792441503 Mon Oct 19 21:25:03 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID 6 SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC 18e16482cd3533TIME 1792451758 Tue Oct 20 00:15:58 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC 18e4d94e9a1451TIME 1792452909 Tue Oct 20 00:35:09 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1Hardware event. This is not a software error.MCE 0CPU 7 BANK 0 TSC 194008a367af59TIME 1792483269 Tue Oct 20 09:01:09 2026MCG status:MCi status:Corrected errorError enabledMCA: Internal parity errorSTATUS 9000004000050005 MCGSTATUS 0MCGCAP c13 APICID e SOCKETID 0MICROCODE 58CPUID Vendor Intel Family 6 Model 141 Step 1root@CyrusNAS5:~#It looks like this could be a thermal issue, on core 7.need to open up and check for dust, mebe redo the thermal paste, relocate to cooler area. Edited October 20Oct 20 by einsteinagogo
October 20Oct 20 These show repeated corrected internal-parity errors in bank 0, mostly on logical CPU 7, with one on CPU 3. “CPU 7” is a logical processor number, so it does not necessarily mean physical core 7.Checking for dust and improving airflow are reasonable steps, but these records do not establish overheating. Please record CPU temperatures and any thermal throttling during normal use, then see whether new machine-check errors appear after improving cooling.There are also entries throughout late September and early October, so the hardware errors continued during the period when the server otherwise appeared stable.If the errors continue at normal temperatures and stock settings, I would seek hardware-vendor diagnosis. These corrected events do not prove that the CPU is defective or that they caused the Btrfs corruption, but they deserve investigation.
October 20Oct 20 Author I establishedThe whole CPU throttled about 3 million times in 25 days, roughly once every 0.7 seconds on average. It’s hitting its 100°C limit constantly, at least in short bursts. The 67–71°C is at light load, with Docker and VMs stopped.CPU 7 throttled about 400 times more than CPU 0, and it’s the core with four of the five errors.Time to check. a quick Google reveals issues with this Intel NUC extreme. Edited October 20Oct 20 by einsteinagogo
October 20Oct 20 Author Coming back to this, no dust to speak of, but I think the Cooling Profile was too low for this NUC. So I've temporary maxed the fans out, disabled TurboB oost, and Max Performance, its not as if I'm using it for Gaming.Tempertatures are now 41 degrees C, so a massive 26 degrees C - change
October 20Oct 20 See if the MCEs stop after this; if they continue to occur, there's likely still some hardware issue.
October 20Oct 20 Author We've remembered that all syslogs go to a central catch all Syslog server before being forwarded to Splunk.So we actually have all the data.It does look like CPU errors, which could be thermal throttling, or overheat.We will monitor over next 7 days with existing current cooling profile.Thanks for your assistance.
October 21Oct 21 Author Stress testing, Seller and ASUS have been contacted, the unit is under warranty with 10 months remaining.for anyone else following the thread, we have re-seated the memory.and running memory tests in UnRaid.We''ve set fans to 100% full, and turned off Turbo.sensors | grep -i package; grep -c "^MCE" /var/log/mcelog; free -h | head -2Package id 0: +53.0°C (high = +100.0°C, crit = +100.0°C)0total used free shared buff/cache availableMem: 62Gi 26Gi 34Gi 25Gi 27Gi 35Gia256sum /tmp/st$t.bin | cut -d' ' -f1); n=$((n+1)); [ "$s" != "$R" ] && echo "$(date +%T) thread $t run $n MISMATCH"; done; echo "thread $t done: $n runs" ) & done; wait Edited October 21Oct 21 by einsteinagogo
October 21Oct 21 Author Test 1 passed. All 16 threads finished, with no MISMATCH lines:about 480 passes per thread, roughly 7,700 checks in totaleach pass read 1.5GB, so about 11TB of data was checked0 CPU errors, at 50–53°C under full load
October 21Oct 21 Author The 8-hour test passed.All 16 threads finished, with no MISMATCH lines. That’s about 1,920 passes each, roughly 30,700 checks in total.With 1.5GB per pass, that’s about 46TB of data checked.0 CPU errors, steady at about 50–53°C.So we've been soa tested and stress tested all day after reseating ramstill need to work on lowerin the fan speeds
October 22Oct 22 Author So the testing continues.When we started to verify data between NVMe storage devices, we started to observe issues in the compare, and then later the kernel crashed!So despite reseating the RAM, which passed the above tests, we started to look at individal SODIMM sticks. Removing and testing individually.Stick A completed 2,5 runs and failed, and crashed the server.Stick B has completed to date 140 runs, with no issues, no compare mismatches, and no kernel reboots.So it looks like with been plague with thermal issues with CPU - possible fix - increase fan speed. (cause of MCE errors)Stick A - possibly faulty?
October 22Oct 22 Author Yes Memtest86+ didn't find any errors, and even when I tested myself for 24 hours, it didn't show any issues inside the OS.Only when started copying large files, I could see the issue.
October 22Oct 22 Author No clear evidence, so I'm going to boot this server, using Ubuntu from a USB stick, and thensudo apt update && sudo apt install -y stress-ng sudo stress-ng --vm 8 --vm-bytes 75% --vm-method all --verify -t 3h --metrics-brief 2>&1 | tee ~/stress-stickB.txtthis then excludes UnRAID, NVMe devices, and just tests the a single stick of RAM and Ubuntu OS, if that crahses then its hardware.then move into to check a Samsung module which is certified for this device, current modules are Kingston Fury modules tested by Kingston for this Nuc.
October 23Oct 23 Author I'm going to open a new Forum post. On Testing a computer for use with UnRaid. Edited October 23Oct 23 by einsteinagogo
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.