9 hours ago, trurl said:

we need diagnostics taken when the problem occurs.

If you can't get diagnostics for some reason, the result of this command line might tell us something:

df -h /

the parity rebuild completed, but the array still showed the drive as invalid so I'm trying it one more time.

tower-diagnostics-20260329-1116.zip

1 hour ago, lockrob2000 said:

parity rebuild completed

Parity rebuild starts here:

Mar 29 04:52:07 Tower kernel: md: recovery thread: recon P ...

End of syslog is here:

Mar 29 05:02:06 Tower root: Fix Common Problems: Warning: Syslog mirrored to flash

With no indication that rebuild completed, and obviously not enough time for that anyway.

Did you take these diagnostics just after starting the rebuild?

Post new diagnostics

Yes, it was after starting the rebuild.

I will do another diagnostic now. This rebuild has about an hour and a half.

Thanks-

tower-diagnostics-20260329-2126.zip

Your syslog is being spammed with these

Mar 29 04:56:10 Tower kernel: [drm] PCIE GART of 256M enabled (table at 0x000000F400900000).
Mar 29 04:56:10 Tower kernel: [drm] UVD and UVD ENC initialized successfully.
Mar 29 04:56:10 Tower kernel: [drm] VCE initialized successfully.
Mar 29 04:56:10 Tower kernel: amdgpu 0000:1c:00.0: [drm] Cannot find any crtc or sizes
Mar 29 04:56:16 Tower kernel: [drm] PCIE GART of 256M enabled (table at 0x000000F400900000).
Mar 29 04:56:16 Tower kernel: [drm] UVD and UVD ENC initialized successfully.
Mar 29 04:56:16 Tower kernel: [drm] VCE initialized successfully.
Mar 29 04:56:16 Tower kernel: amdgpu 0000:1c:00.0: [drm] Cannot find any crtc or sizes

and it repeats like that about every 3 seconds.

Any idea what that is about?

Sorry, I don't know what any of that means.

Not even sure where to start looking.

Besides my GPU, I have a PCIE SATA expander and a PCIE network card. I could remove both of those if it would help.

The SATA card is for moving some of the hard drives out of the case because i was worried heat issues might be causing the disconnects, and the network card was just installed because I was worried about the network card perhaps being an issue.

What do you think?

Edited March 30Mar 30 by lockrob2000

Other than that nothing seems to be wrong, rebuild hasn't completed yet.

So what exactly do you mean by "disconnects"? Do you mean your disks are disconnecting. Or do you mean you can't access the webUI?

9 minutes ago, lockrob2000 said:

What do you think?

I think you are making your hardware unnecessarily complicated, difficult to maintain and troubleshoot.

I know you probably know this but

1. What does it mean?

• Successful Hardware Setup: The system has successfully loaded the hardware acceleration capabilities for AMD graphics cards, specifically for video decoding (playing) and encoding (recording/transcoding).

• AMDGPU Driver: This is a normal log entry for AMD Radeon GPUs (especially Polaris/RX 400/500 series and newer) during boot or when a Virtual Machine/Docker container loads.

• VCE Initialized: It is frequently followed by "[drm] VCE initialized successfully," which refers to the Video Coding Engine.

  Unraid +3

2. Is this an error?

• Generally, No: In most cases, this is an informative message indicating that the hardware is ready to work.

• If it spams the logs: If this message appears every 10–30 seconds in your syslog, it indicates a "log spam" issue, often related to a known AMD reset bug where the card is not correctly resetting, but it is not necessarily crashing your server.

3. Contextual Usage in Unraid

• Docker/Transcoding: If you use Plex, Jellyfin, or Frigate with an AMD GPU for hardware transcoding, this message confirms the driver is loaded and ready to process video.

• VM Passthrough: If you pass through an AMD GPU to a Windows VM, this means Unraid has initialized the card, though it may still suffer from the "AMD Reset Bug" (failing to reset on VM shutdown).

  Unraid +3

4. What to do if it's spamming logs

If this message is filling up your log file (causing the log to reach 100%), it usually means the AMD GPU is being initialized repeatedly.

• Install the vendor-reset plugin: For AMD GPU passthrough users, this plugin can help address the reset bug that causes these logs.

• Check for Idle Issues: If the GPU is unused, it may be going into a low-power state and re-initializing.

  Unraid

I have been grasping at straws trying to find out what is causing these issues.

Sorry, what I mean by disconnects is that when used to work with media, the server will works for a period of time (generally around an hour)_ and lose connection to my network. I can't tell what it actually does, but I can no longer connect in any way, even though the power light still shows "running" while the network router, shows the server as off.

To get it to reconnect, I have to power the server off using the power button, and after a few seconds power it back on.

It will then reconnect for another indeterminate period of time. Sometimes as short as fifteen minutes, sometimes hours.

5 minutes ago, lockrob2000 said:

I know you probably know this but

No, I don't use AMD and didn't research this. The mentioned plugin can be installed from the Apps page.

Post a new screenshot of your Syslog Server settings.

Just took this.

OK, that looks good for syslog server. You might consider turning off Mirror to flash now since it will be saving to your Photos share.

Next time you have the "disconnect", you can get the syslog from your Photos share, zip it, and post it.

Just now, trurl said:

You might consider turning off Mirror to flash now since it will be saving to your Photos share.

And enable syslog rotation so the files don't get too large.

It's OK to leave Syslog Server enabled if it isn't writing to flash, you can just delete the logs manually after a while.

What's a good size to set the maximum file size to?

When we are trying to see what happened leading up to the server becoming unresponsive, we usually don't need a very big file. 1MB is probably plenty, and maybe only keep 1. You can do more if you want, it won't be using a lot of your storage capacity.

Please zip before posting though. It is text and will compress very well.

It seems to have worked.

now I just have to do the tweak the C-state, remove the added PCIE additions, I've added the AMD vendor reset already.

Fingers crossed.

I really appreciate your help and patience.

Thanks-

Edited March 30Mar 30 by lockrob2000

Just in case you want to see this. It starts in February for some reason, but does include up until tonight.

syslog-127.0.0.1.zip

OK, my instance has been running for 18.5 hours with no disconnects.

I can surmise that the issue was the C-states settings. I went into BIOS and disabled C-states and that seems to have been the golden ticket.

Thanks you for all the help. I was thrashing around blindly until you helped steer me in the right direction.

Thank you so much!