My setup:

·         Ryzen 5 3600

·         ASRock B550 Pro4

·         Corsair Vengeance LPX 32 GB (2x16)

·         HP EX900 250 GB M.2-2280 NVME (unassigned disk)

·         Samsung 860 EVO 500GB (cache)

·         Ironwolf NAS 4 TB drives (x6)

·         Shucked WD MyCloud 6 TB drive (parity)

·         ZOTAC GeForce GTX 1070 Mini

·         MSI GeForce GTX 1050 Ti

·         Ziyituod PCIe SATA Card

·         SeaSonic M12II 520 Bronze EVO Edition 520W (I understand it may be slightly underpowered running at full load, but I was having the same stability issues while only using the 1050 ti)

·         Running Nvidia 6.8.3 from unraid nvidia plugin

 

Plugins:

·         Nvidia unRAID

·         Fix Common Problems

·         Dynamix SSD Trim

·         Community Applications

·         Tips and Tweaks

·         Unassigned Devices

·         Unassigned Devices Plus

 

Docker Containers:

·         Plex Media Server

·         Sonarr

·         Radarr

·         NZBget

 

Issues:

Server becomes unresponsive requiring a hard reboot during, seemingly, any task. Invoking mover, Plex library scan, nzbget downloads, using krusader, parity check, copying files from the server to my laptop, even just letting the server just sit with nothing but the array running will eventually end in a crash. The thing is none of these events cause it to crash instantly. I can start a Plex library scan, for example, and let it run for maybe 10 minutes before everything becomes unresponsive. Sometimes invoking the mover causes the system to lock up, and other times it finishes with no issues.

 

What I’ve tried:

·         Disabled C-States in BIOS

·         Set Typical Current Idle

·         Replaced all SATA cables

·         3 different RAM sets

·         Memtest for 36 hours (0 errors)

·         Swapped MSI X570A Pro for ASRock B550 Pro

·         Formatted flash drive saving only license file

·         Removed all docker containers and plugins

·         Deleted all appdata, docker.img, VMs

·         Added “rcu_nocbs=0-11” to syslinux config

·         Checked and double-checked container paths

·         Checked SATA and power connections every time I dove into the case

 

My end goal is to use the system for Plex streaming with hardware transcoding and media downloading, using the VMs as daily drivers accessed through RDP from my Chromebooks for myself and kids (online school starting up), and some light gaming with the passthrough GPU.

 

At this point, the only things I haven’t tried to replace are the CPU, flash drive, and power supply. I am looking at a 750W PS due to the dual GPU anyway, but the current PS should be fine (bought brand new and the server is never at max load). I have a Ryzen 3 3100 on the way to test the CPU and I'll be trying a new flash drive soon. I have been Googling issues and attempting fixes for months now. I can’t see any smoking gun in the syslog. Hoping someone here can shed some light into the issues I’ve had with this build.

 

 

tower-diagnostics-20200908-1847.zip

Edited September 10, 20205 yr by belanj89
additional info

Problem seems to be solved with the new CPU. Everything is stable. I didn't want to believe a new processor could be bad, but I guess it was.

Several crash report ( Ryzen 3000 series CPU ) seems not resolve, it may be good ref. if you identify it is CPU problem.

 

Does more CPU info. can provide, i.e. stepping, date code, etc

On 9/12/2020 at 1:34 AM, belanj89 said:

Problem seems to be solved with the new CPU. Everything is stable. I didn't want to believe a new processor could be bad, but I guess it was.

Your journey sounds very much like the one that I'm about 14 days into 😫  Glad you got sorted. 👏

 

My setup is:

• ASRock X570M Pro4
• AMD Ryzen 9 3900X
• CRYORIG C7 Cu - cpu fan
• 2 *  Crucial 16GB DDR4 DIMM 2666 MHz / PC4-21300 ECC (CT16G4WFD8266)
• SilverStone Technology CS381B
• LSI SAS 9207-8i Host Bus Adapter Kit (8 Port Internal, 6Gb/s SATA+ SAS, PCIe 3.0 HBA)
• be quiet! BN639 600W SFX L Power Supply
• 2 * Corsair CSSD-F240GBMP510 Force Series MP510 240 GB NVMe PCIe Gen3 x 4 M.2 Solid State Drive - (cache drive pool)
• 2 * SamTones Internal Mini SAS SFF-8087 to Mini SAS High Density HD SFF-8643 Data Server Hard Disk Raid Cable 50cm
• MSI GT 710 1GD3H LPV1 - (graphics)
• 6 * 12tb Seagate ST12000vn0008 Ironwolf NAS hard drives (data array)
• CruzerFit - 64GB Flash Drive
• UnRAID 6.9.0
   

What I’ve tried:

• Flashed X570M Pro4 with latest v3.40 BIOS rom
• Disabled C-States in BIOS
• Set Typical Current Idle
• Memtest for 24 hours (0 errors)
• PassMark Burin Test for 24 hours (0 errors), with a "sacrificial" SATA drive in each of the bay stacks, so that I could run the disk tests and exercise the HBA and SAS interconnects
• Added “rcu_nocbs=0-23” to syslinux config
• Increased both chassis fan speed to FULL in BIOS
• Increased CPU fan speed to FULL in BIOS

However the Memtest and Passmark tests took a few attempts before completing successfully.  The interesting thing is that Memtest was a linux based boot USB, where as Burnin in WinPE based boot USB, both were initially problematic.

 

My next move today is a total hardware tear-down, and rebuild and retry Unraid. If that fails I'm going to install Windoze on a spare SATA drive to hopefully determine if this is a Ryzen/Linux thing, or a pure hardware thing.

 

I'm half way through my UNRAID trial, and the longest continuous time that box has stayed up in about 3, maybe 4 days!!!

 

Factoring in, the time, effort, and general loss of sanity this build has cost me, I'm starting to wish I'd just bought the QNAP!  Sharing this as I've found it useful to read other people's experiences.

 

Good luck fellow Ryzen users. 🤞