Unraid kernel panic (Brand new machine)

[sloob]

Hello everyone.

 

I recently acquired a brand new machine to run unraid.

specs are as follows;

 

MOBO: ROG STRIX B550-F GAMING (WI-FI) (Most recent BIOS version)

CPU: AMD Ryzen 5 5600X 6-Core @ 3700 MHz

HBA card: INSPUR 9211-8i 6Gbps HBA LSI FW:P20 IT Mode ZFS FreeNAS unRAID+2* SFF-8087 SATA

RAM 32GB Corsair Vengeance RGB PRO DDR4 3200MHz

PSU: Corsair gold 850W

Drives: Western Digital Red 6TB

Cache: Samsung 970 EVO Plus 1TB NVMe M.2 Internal SSD (MZ-V7S1T0/AM)

Flash drive: SanDisk 32GB Cruzer Fit USB Flash Drive - SDCZ33-032G-G35

 

Everything runs well but about once a week we get a kernel panic crash. I ran the mem test for a few hours and got no issues.

The server is plugged in a working UPS.

 

Can anyone figure this out?

 

jarvis-diagnostics-20220322-1213.zip

[sloob]

Anyone?

This is for our office and I kinda put my neck on the line and it keeps crashing.

 

any help appreciated

[JonathanM]

 

[Hoopster]

On 3/24/2022 at 7:02 AM, sloob said:

Everything runs well but about once a week we get a kernel panic crash

Have you checked out the suggestions in this post related to Ryzen-based servers?  Memory overclocking (pay attention to what speed the board/CPU can support and not the rated speed of the memory) and C-states seem to be the most common causes of Ryzen-based system issues.

Edited March 25, 2022 by Hoopster

[sloob]

2 minutes ago, Hoopster said:

Have you checked out the suggestions in this post related to Ryzen-based servers?  Memory overclocking (pay attention to what speed the board/CPU can support and not the rated speed of the memory) and C-states seem to be the most common causes of Ryzen-based system issues.

Sorry I completely forgot to mention this in my post. Yes C state is disabled and the server is running at 3200 MHZ. should I try to lower it even tho it is supported by my motherboard?

[Hoopster]

8 minutes ago, sloob said:

Sorry I completely forgot to mention this in my post. Yes C state is disabled and the server is running at 3200 MHZ. should I try to lower it even tho it is supported by my motherboard?

Just make sure the speed at which the RAM is running (as reported in the BIOS) is equal to or lower than the speed the CPU can support according to the chart in the linked post.  Also note that supported RAM speed is often dependent on how many RAM slots are filled.  XMP profiles are often overclocks.

[ChatNoir]

8 minutes ago, sloob said:

the server is running at 3200 MHZ

Not sure about that, the diagnostics show 2133, as if the XMP profile is not applied. So the issue is probably not cause by memory speed.

[sloob]

1 minute ago, ChatNoir said:

Not sure about that, the diagnostics show 2133, as if the XMP profile is not applied. So the issue is probably not cause by memory speed.

Shoot, that would have been an easy fix! Can you think of anything else?

[ChatNoir]

30 minutes ago, sloob said:

Shoot, that would have been an easy fix! Can you think of anything else?

Not without more information. The regular logs are stored in RAM and are reset on boot. You should setup a syslog server and post the file after a crash.

[sloob]

59 minutes ago, ChatNoir said:

Not without more information. The regular logs are stored in RAM and are reset on boot. You should setup a syslog server and post the file after a crash.

I did enable mirror to flash before the last crash. would this file be of any help?

syslog

[ChatNoir]

8 hours ago, sloob said:

I did enable mirror to flash before the last crash. would this file be of any help?

Your call trace mentions MACVLAN. It might be related to containers with custom IP.

You should update to the latest 6.10 RC and change your docker network type to IPVLAN.

 

 

[sloob]

5 hours ago, ChatNoir said:

Your call trace mentions MACVLAN. It might be related to containers with custom IP.

You should update to the latest 6.10 RC and change your docker network type to IPVLAN.

 

 

Thank you! I will change the MACVLAN to IPVLAN in the docker settings (I had already done that actually, not crashes yet) and report back!