The UnRAID server unexpectedly stops working.
The UnRAID server unexpectedly stops working.
I'm getting desperate... Since I built this NAS, I've been having random crashes that I am unable to the cause. When I built the server, I had trouble to get it to post and the CPU and RAM error lights would turn on. After reseating those components it was able to post, but that problem was back after moving the server into the rack. It started working after a few power cycles and was stable for a while. After a 1-2 months, the system started crashing maybe once every few weeks. Since around 2-3 months ago it started crashing every week. In the past week it has been crashing every 12 to 36 hours. It never occurs at the same time of the day. None of the crashes give me an error code or leave traces in the syslog. No errors are displayed on the server's monitor (if there is, I don't have time to look at it since I don't sit in front of that monitor all day long to see it). Every time, the system immediately reboots just like if I pressed the "reset" button. I've created a post on the UnRAID forums a while back, but they were not able to give me clues for a cause. However they seem certain it is not caused by UnRAID. You can look at the post here: https://forums.unraid.net/topic/155746-r...ent-cause/ Here's what I tried so far, though I might forget some things so let me know if you need any additional info: 1. Update BIOS to latest versions 2. Replace CPU with a new one (Intel 13th Gen RMA) 3. Check OS for defects 4. Reseat power cables 5. Reseat RAM 6. Reseat SATA cables 7. Upgrade CPU cooler 8. Add case fans to improve air flow 9. Check if UPS is faulty/Change UPS None of these helped. Changing the CPU however, made me optimistic after 1 week of no crashes. That optimism what short-lived. Here's my parts list: Motherboard : AsRock Z790 Pro RS WiFi Bios version : 13.01 CPU : Intel i7 13700k Memory : Crucial Pro RAM 32GB Kit (2x16GB) DDR5 5600MHz Power supply : EVGA 600 B1, 80+ Bronze 600W PCI-E device : HP H220 6Gbps SAS PCI-E 3.0 HBA LSI 9207-8i P20 (in IT Mode) HDDs : 3x 10TB WD Red, 1x 8TB WD Red, 3x 4TB WD Red SSDs : 2x 500GB SATA (cant remember the brands) Only the HBA was bought second hand, but the issue was present before installing it. The power supply was used in my previous NAS and had no issues. This system was built in december 2023. As a bounty to help me fix this, I offer pictures of my cat. Thanks in advance
Hey, start with one stick on RAM and switch to the other. Unriad connects straight from RAM, so issues there or instability could cause problems. Turn off XMP settings if it's enabled. If it freezes when idle, run the server without an HBA and no storage.
I plan to swap out a RAM stick after work. I verified XMP is enabled automatically (Load XMP Setting: Auto) and can’t turn it off. I should note that Disk 1 (8TB) was disabled in two separate incidents—once a month ago and again this Monday. The previous 4TB drive began showing read errors in S.M.A.R.T. data, causing file system corruption and slowing it down to 1.5GB/s. During an extended self-test, the process halted repeatedly and at one point the disk was reassigned to unassigned devices. Now it’s back to normal speed but remains disabled; I’ll need to re-add it and re-sync parity. I ran memtest86 for about 8–9 hours, completing all 11 passes without issues.
Up until now, I experienced just one crash shortly after replacing the RAM module. I've had sufficient time to reinitialize an inactive drive and keep up with updates for unraid. I plan to watch closely for a few days before alternating the RAM swaps again to verify the problem.
Ensure you're using the latest UNRAID release. Older versions may encounter macvlan problems.
I'm on December 6th, 2013, and before that it was December 6th, 2011.
I've experienced two additional crashes within the past 12 hours. I'm attempting to gather more system data to identify potential causes. In Netdata, the voltage sensor shows some devices shutting down gradually before the crash occurs. This pattern isn't visible in the metrics collected via Prometheus, where all indicators—memory usage and CPU temperature—remain stable. Unraid only provides limited metrics, while Prometheus runs on a separate machine.
Brother, it seems I found the fix I was looking for. I didn’t realize Unraid used RAM in that way. After installing more RAM, I faced boot issues and crashes. I plan to turn off XMP when I’m not playing games with players and try again. I’ll let you know what happens!