Hi All,

Wonder if one of you can help me better than the Unraid support rep I've had...

Basically, my system is an unstable mess right now, and I have no idea where to start processing this. It's suffering from a complete system freeze, with no response to anything (Web UI, SSH, Ping). I have to pull the power to bring the system back up. The crashes are random. Seemingly appearing when no one is using the system at that moment or when people are using it. It's at least once a day

I've tried different USBs and other ports. I can run Proxmox fine with zero issues like this. But Unraid is just too unstable for my stack at the moment.

I'm on the latest version 7.2.2. This is a fresh install with a handful of *Arr apps, Plex running with Nvidia drivers, and Unassigned devices just for NFS. Occasionally, I see the odd SegFault error in the kernel log. But it doesn't seem to impact anything. I've run scans on the NVME's in the machine, and they've passed SMART and other checks (About 10% worn)

The machine is a Dell Precision 3280 Compact with an Intel Core i9-14900, 64GB RAM, and Nvidia T1000 8GB, which has passed all Hardware Diags in the BIOS.

Any ideas for me to look into would be massively appreciated, as I'm starting to pull my hair out. 😥

Thanks!

tower-diagnostics-20251125-1032.zip

Start by enabling the syslog server, then post that after the next crash.

So my server crashed this morning. Hardlocked up, no way to connect unless I pull the power. Checking DMESG, I see SQUASHFS errors.

This is a brand-new SanDisk USB (about 1 month old), and I also used a Samsung USB before then. I'm aware of the issues with SanDisk USBs on this forum. But I don't believe this is the issue, since I've used two different USB sticks (originally three), and I've seen the problems with my Samsung USB. So would that point to a problem with the machine's USB ports? This machine has no problem when the OS is installed to an NVME drive (Proxmox OK, Debian 13 OK, Truenas OK)

[   48.238604] SQUASHFS error: xz decompression failed, data probably corrupt
[   48.238607] SQUASHFS error: Failed to read block 0x21835aa4: -5
[   48.238609] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   48.238613] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   48.238613] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
[   48.789518] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   48.789522] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
[   49.350254] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   49.350258] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
[   49.904344] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   49.904348] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
[   50.461711] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   50.461714] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
[   51.026955] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   51.026959] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
[   51.590774] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   51.590778] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
[   52.148430] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   52.148434] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
[   52.703884] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   52.703888] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
[   53.259070] SQUASHFS error: Unable to read fragment cache entry [21835aa4]
[   53.259074] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90

Added syslog. It might be helpful, but I can't see anything that stands out.

2025-11-25T07:05:44+00:00 Tower kernel: traps: php-fpm[553878] general protection fault ip:941736 sp:7ffd3d8c5140 error:0 in php-fpm[541736,600000+3fb000]
2025-11-25T07:05:44+00:00 Tower nginx: 2025/11/25 07:05:44 [error] 242977#242977: *15236 recv() failed (104: Connection reset by peer) while reading response header from upstream, client: 10.0.0.59, server: , request: "POST /plugins/community.applications/include/exec.php HTTP/1.1", upstream: "fastcgi://unix:/var/run/php-fpm.sock:", host: "10.0.0.66", referrer: "http://10.0.0.66/Apps"
2025-11-25T10:59:13+00:00 Tower sshd-session[170715]: Read error from remote host 10.0.0.59 port 63339: Connection timed out
2025-11-27T02:27:21+00:00 Tower kernel: SQUASHFS error: Unable to read fragment cache entry [21835aa4]
2025-11-27T02:27:21+00:00 Tower kernel: SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
2025-11-27T02:27:21+00:00 Tower kernel: SQUASHFS error: Unable to read fragment cache entry [21835aa4]
2025-11-27T02:27:21+00:00 Tower kernel: SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
2025-11-27T02:27:22+00:00 Tower kernel: SQUASHFS error: Unable to read fragment cache entry [21835aa4]
2025-11-27T02:27:22+00:00 Tower kernel: SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
2025-11-27T02:27:23+00:00 Tower kernel: SQUASHFS error: Unable to read fragment cache entry [21835aa4]
2025-11-27T02:27:23+00:00 Tower kernel: SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
2025-11-27T02:27:23+00:00 Tower kernel: SQUASHFS error: Unable to read fragment cache entry [21835aa4]
2025-11-27T02:27:23+00:00 Tower kernel: SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
2025-11-27T02:27:24+00:00 Tower kernel: SQUASHFS error: Unable to read fragment cache entry [21835aa4]
2025-11-27T02:27:24+00:00 Tower kernel: SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
2025-11-27T02:27:33+00:00 Tower root: status : errored
2025-11-27T02:27:33+00:00 Tower root: error log : /var/log/graphql-api.log

1 hour ago, Threshermaw said:

[   52.703888] SQUASHFS error: Unable to read page, block 21835aa4, size 6d90
[   53.259070] SQUASHFS error: Unable to read fragment cache entry [21835aa4]

This is typically a bad flash drive, or bad hardware, most often RAM or the CPU.

Even if it was replaced recently, it may be worth doing it again to make sure, to test the RAM, and since you have two DIMMs, you can try with just one, if it still crashes, try the other one, that will basically rule out the RAM.

The final suspect would be the CPU, since it's a 14900 it could be affected by the Intel 13/14 gen issues. Look for a BIOS update; it can help if the CPU is not too damaged already.

24 minutes ago, JorgeB said:

This is typically a bad flash drive, or bad hardware, most often RAM or the CPU.

Even if it was replaced recently, it may be worth doing it again to make sure, to test the RAM, and since you have two DIMMs, you can try with just one, if it still crashes, try the other one, that will basically rule out the RAM.

The final suspect would be the CPU, since it's a 14900 it could be affected by the Intel 13/14 gen issues. Look for a BIOS update; it can help if the CPU is not too damaged already.

I have now removed one stick, running 32GB, to see if the problem improves.

Currently running the latest (1.18) BIOS for this machine. Will update.

Problem with USB changes is that my previous one is blacklisted and i don't have another spare USB hanging around... Required a ticket to Unraid Support to fix..

2 hours ago, Threshermaw said:

Required a ticket to Unraid Support to fix..

Automatic replacement works twice per year, only after that do you need to contact support.

Update - So. After another crash, I've decided to disable Intel Turbo Boost on my CPU, and the system is FAR more stable (believe this might be a voltage-damaged 14th gen). I'm going to run this for a few days to see if we get a better experience.

Final - It appears to have been CPU-related. The machine had been up for 5 days with no hard crashes. I got Dell to send me a new 14900 CPU to replace this once, and I have now replaced the CPU and re-enabled TurboBoost, and all seems well—no SegFaults in the logs.

Leaving this here to help another fellow homelabber work out what the problem could be. It is more than likely a hardware issue. Contact your warranty provider.