Unraid 6.7.2 - Freezes Overnight - MCE Hardware Errors

[Arizuia]

Hello all you fine fellas.

I've recently upgraded my Unraid machine from old chunky Q6600 machine to a mATX build. Some specs:

CPU: Ryzen R5 1400 @ stock
RAM: 16GB DDR4 @ 3200MHz
MOBO: Gigabyte B450 Aorus M
PSU: Seasonic 550W (Good Tier)

 

The server is used for NAS and running servers on a Linux VM.

 

First night the server was fine, but the other night it froze and was unresponsive, didn't even let me type in the terminal. Same thing happened next night.

I installed Fix Common Problems plugin and it reported a MCE Hardware Error, I then looked into the log and found this:

mce: [Hardware Error]: Machine check events logged
mce: [Hardware Error]: CPU 0: Machine Check: 0 Bank 5: bea0000000000108
mce: [Hardware Error]: TSC 0 ADDR 1ffff816560ea MISC d012000100000000 SYND 4d000000 IPID 500b000000000
mce: [Hardware Error]: PROCESSOR 2:800f11 TIME 1571423384 SOCKET 0 APIC 0 microcode 8001138

 

Seems to be either CPU or RAM issue, but the same components were in my main system for about a year before I upgraded them and I had no issues with them during that time. Seems odd.

 

I thank anyone who can help me with this.

[trurl]

Have you done memtest?

 

Setup Syslog Server:

 

https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=781601

 

and get us the diagnostics plus whatever syslog captures before the crash.

 

Tools - Diagnostics, attach complete diagnostics zip file to your next post.

[Squid]

3 hours ago, Arizuia said:

it froze and was unresponsive,

 

3 hours ago, Arizuia said:

Ryzen R5 1400

Make sure that C-States are disabled in the BIOS

[John_M]

4 minutes ago, Squid said:

Make sure that C-States are disabled in the BIOS

Actually, the better solution is to look for a "Power supply idle mode" setting in the BIOS* and set it to "Typical current idle" rather than the default "Auto". That still allows the CPU to enter C states but doesn't allow the power to drop so low that it can't wake up again. The issue only affects 1000-series processors.

 

*Typically under Advanced -> AMD CBS -> Zen Common Options

[Arizuia]

3 hours ago, trurl said:

Have you done memtest?

 

Setup Syslog Server:

 

https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=781601

 

and get us the diagnostics plus whatever syslog captures before the crash.

 

Tools - Diagnostics, attach complete diagnostics zip file to your next post.

 

I've left memtest running for a while, at 3200MHz OC there were quite many errors, over 700. Removing OC reduced that to 2 so far.

Attaching the diagnostics file as you asked.

kotiservu-diagnostics-20191018-1834.zip

[John_M]

Don't run the CPU's memory controller beyond its spec. You have to set up a server differently from a gaming machine. Two memory errors is a fail.

[Arizuia]

On 10/19/2019 at 5:08 AM, John_M said:

Don't run the CPU's memory controller beyond its spec. You have to set up a server differently from a gaming machine. Two memory errors is a fail.

After reseating the RAM and keeping it stock, everything seems to be fine now. No memtest errors and hasn't crashed or frozen during last night.

 

Very late update: I bought an used Ryzen 3200G and it solved everything, seems the Ryzen 1400 has some early Zen bugs regarding Linux and stability.

Edited April 7, 2021 by Arizuia
late update