Hi, my Unraid has found some errors, specifically one that shows that I am out of memory. Can someone please take a look at my diagnostics and advise what I need to do to fix it. MANY THANKS! My son is my admin but he is not available to help.

tower-diagnostics-20251130-1642.zip

Using Gemini, it gave me the following instructions to solve my issue. It seems to have worked. Pasting it below in hopes it will help others.

Yes, I can see exactly what is causing the issue in your syslog.txt.

The "Out of Memory" error is a symptom, but the root cause is your NVIDIA GPU driver.

The Problem

Your log is being flooded with millions of error messages from the NVIDIA driver (NVRM). This is filling up your server's RAM (because Unraid stores logs in RAM), which triggers the "Out of Memory" warnings.

Here are the key lines from your file:

• Nov 30 04:40:01 Tower kernel: NVRM: kbifInitLtr_GB202: LTR is disabled in the hierarchy (Repeated hundreds of times)

• Nov 30 05:53:56 Tower kernel: NVRM: nvCheckOkFailedNoLog: Check failed: Out of memory [NV_ERR_NO_MEMORY]

What this means: The code GB202 indicates you are using a very new NVIDIA GPU architecture (likely the RTX 50-series you mentioned previously). The driver is struggling to communicate with the card's power management features (LTR = Latency Tolerance Reporting) and is failing to initialize the adapter, getting stuck in a loop that eats up all your memory.

How to Fix It

You need to stop the driver from spamming the logs and fixing the communication issue.

Step 1: Update the Nvidia Driver

1. Go to the Settings > User Utilities > Nvidia Driver plugin in Unraid.

2. Check if there is an update available. You likely need the very latest "Production" or "New Feature" branch (e.g., version 580.x or newer) to support the RTX 5060 Ti properly.

3. If you are on an older version (like 565.x or 570.x), update it and reboot.

Step 2: Modify Boot Parameters (Crucial for the "LTR" error) The "LTR is disabled" error is a PCIe power state issue often seen on newer cards with certain motherboards. You can force it off.

1. Go to the Main tab in Unraid.

2. Click on Flash (the name of your USB drive) to open its settings.

3. Scroll down to Syslinux Configuration.

4. Look for the line starting with append. It usually looks like: append initrd=/bzroot

5. Add pcie_aspm=off to the end of that line. It should look like this: append initrd=/bzroot pcie_aspm=off (Note: Ensure there is a space before the new command).

6. Click Apply.

Step 3: Reboot Restart your server. This will:

1. Clear the full log file (freeing up your RAM).

2. Apply the new driver and boot settings.

Once rebooted, check your log again. The NVRM spam should be gone, and your memory usage should return to normal.

13 hours ago, ag3n7 said:

The "Out of Memory" error is a symptom, but the root cause is your NVIDIA GPU driver.

Sort of, while it was the nvidia card falling over, it didn't cause your system to go OOM (it would take and awful lot of log to fill up available RAM and knock the host over). AI is great, but it only ever gets it sort of correct.

FCP saw the string 'out of memory' in the log and spit out the warning.

Nov 30 05:53:56 Tower kernel: NVRM: nvCheckOkFailedNoLog: Check failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from _memdescAllocInternal(pMemDesc) @ mem_desc.c:1359

You need to update the GPU driver and/or motherboard BIOS - if that fails, set it to use a lower PCIE generation in BIOS to see if the errors stop