Hi all,

I am new in this community. I used so far a Synology NAS based on 2 HDD in RAID 1 and I have been happy for more than 10 years with this solution.

Now I decided to move my home NAS to a full NVME solution and for this purpose I bought a GMKTec NucBox G9 pc.

 

Technical Specifications:

CPU: Intel N150

Memory: 12 GB LPDDR5

Network: 2 x 2.5GbE LAN Intel i226V

Storage: 4 x NVME Slot PCIe Gen 3 x2

 

I also took 3 x Kingston FURY Renegade PCIe 4.0 NVMe M.2 SSD of 4 TB each

 

I installed Unraid 7.0.1 and setup a pool with ZFS and RAIDZ1 configuration. So far so good. I started having stability issues when I started transferring data via Network from the Synology to Unraid. When I start this operation, the system crashes and reboot in no more than 1 hour.

 

What I already tried

• Changed the following BIOS Settings:
  • CPU Profile: Quiet, Balance, Turbo (it changes the peak frequency of the CPU)
  • C-State support: tried both disabled and enabled
• Using only one ethernet port at time: same behavior
• Doing memtest: no errors or crashes
• Doing Linpack Xtreme stress test: no errors or crashes
• Putting additional fans to improve temperatures
  • Without additional fans: CPU reach 80 °C and NVME disks reach 60 °C
  • With additional fans: CPU reach 60 °C and NVME disks reach 40 °C
  • However in both configuration I got the same crashes

 

This problem occurs with a very standard configuration Unraid just installed: no Docker, no VMs, nothing of nothing. Only one Share exposed via NFS, mounted on Synology and started the file copy. I also installed some more plugin to better monitor the hardware, but nothing has been useful.

 

In general I was not able to find nothing in the syslog (I enabled them to troubleshoot). Also Unraid doesn't give me any error about the Disks. I am attaching at this post the diagnostics and the last two syslog files between the last crash.

 

Do you have an idea about this issue?

 

Many thanks

 

unraid-diagnostics-20250402-1415.zip syslog-previous syslog

26 minutes ago, vifani said:

the system crashes and reboot in no more than 1 hour.

Server rebooting by itself is almost always a hardware issue, if you have multiple RAM sticks, try with just one, if the same try the other one, that will basically rule out a RAM issue, since memtest is only definitive if it finds errors.

1 hour ago, JorgeB said:

Server rebooting by itself is almost always a hardware issue, if you have multiple RAM sticks, try with just one, if the same try the other one, that will basically rule out a RAM issue, since memtest is only definitive if it finds errors.

Thanks for the answer. Unfortunately I cannot do this test because the RAM is soldered to the motherboard and not expandible in this mini-pc. Do you know some additional mem test I can do to really stress memory and check it? I mean, even to request an RMA I should in someway prove a RAM issue

There's a mem tester plugin, maybe that one detects something, and at least you can run the server while testing.

2 hours ago, JorgeB said:

There's a mem tester plugin, maybe that one detects something, and at least you can run the server while testing.

I will try it. I am also trying to create a Pool with one NVME disk at time and stress it... to understand if maybe one of three drives is faulted even though I found nothing on syslog about nvme errors or smartctl.

I tried the mem tester plugin, but it has not been able to find any issues.

I also tested the following setups:

• zfs with only one nvme ssd: no crashes during high load transfer
• zfs with 2 nvme ssds in mirroring: no crashes during high load transfer
• zfs with 3 nvme ssds in raidz1: crashes during high load transfer after 2 TB of data written

This is really weird and I don't know how to diagnose an hardware problem even to request an RMA 😔

Yep, it's a weird one, suspect board or CPU issue.

I saw that another user had similar issues with a similar configuration (transfer big amount of data via NFS and same Intel NIC). I will try to perform a big data import from an USB drive to check if it can be an NFS/NIC related issue
 

 

Some updates. Yesterday I installed my 4th 4 TB HD (I am planning to do a Raidz2 configuration) and I moved the mini-pc to another place where I used another USB-C power supply I have as spare. Note: this mini-pc uses an USB-C port as power input.

 

In this configuration, the system was rebooting immediately as soon as I click "start array". I was a little bit desperate and I found in some other threads that adding the options "nvme_core.default_ps_max_latency_us=0 pcie_aspm=off pcie_port_pm=off" as kernel boot parameters could help. However also with those parameters the system was rebooting clicking the star array button. So I decided to bring back the mini-pc to the previous location, re-using its original USB-C power supply and magically everything is working without issue since then. I have an uptime of more than 24 hours (first time I reached with this mini-pc) without issues. I have been able to transfer from Synology to Unraid more than 3 TB of data and also my Plex Media Server installation (in Docker) and it's all working without issues.

 

So basically the USB-C power supply used in the new location was totally not able to manage the power required by the mini-pc, but I guess that the options "nvme_core.default_ps_max_latency_us=0 pcie_aspm=off pcie_port_pm=off" fixed the original problem that I think it is caused my some weird behavior of the power management of the PCI Express lines by the motherboard or maybe by the PCI-E Switch controllers (ASMedia ASM1182e) that are in place in this mini-pc.

 

I will monitor the current setup for some days to understand if it is really stable or not.