I am having issues with my UNRAID server losing internet connection every 24-48 hours at night like clockwork. I will start by saying I am new to hosting and built this server about two weeks ago with all new parts. Following some threads of people dealing with this issue online, I found that I was initially on macvlan and switched to ipvlan. This did not help. I then thought maybe my onboard Realtek Ethernet switch was the issue, so I downloaded a driver for it, which did not help. I then tried adding on a 10 G Intel NIC to see if this would help, which it also did not. I tried changing Ethernet cables and ports for sanity checks to no avail.
My thought process left me to believe this was either going to be some type of hardware issue or a BIOS/mobo power setting that I had not appropriately set. My RAM is new, but I did not run a memory test on it. I turned on logging to see what was going on at this point, which is included below.

Environment

• Unraid 6.12.24 on an Intel i5‑12600K system with 32 GB RAM

• Primary interface: Intel X540‑T2 10 Gbps NIC as br0, static IP 192.168.1.252

• Docker set to ipvlan on br0 (initially was macvlan, then switched to ipvlan)

• ZFS pools in use with default ARC (≈50% of RAM)

Symptoms

• Server and dockers become randomly unreachable overnight.

• Only power‑cycling the NAS fixes the issue

• Syslog logs repeated “Link is Down”/“Link is Up” and NTP interface resets, but no persistent NIC error

• Final crash always coincides with massive __vm_enough_memory failures and segfaults in udevadm and zfs processes

Troubleshooting steps

1. Realtek driver instability

   • Moved from in‑kernel r8169 to Realtek r8125 out‑of‑tree driver; reduced flaps but did not eliminate outage

2. NIC hardware swap

   • Installed Intel X540‑T2, confirmed ixgbe driver and sustained 10 Gbps throughput with iperf3; issue persisted

3. Docker networking

   • Read that macvlan/macvtap causes host <-> container conflicts; switched custom networks to ipvlan on br0

4. Network config audit

   • Verified /boot/config/network.cfg only defines br0→eth0, static IPv4/DNS, DHCP off; no leftover vhost0 or macvlan entries

5. Physical & BIOS checks

   • Replaced the cable and switch port

6. Persistent syslog collection

   • Enabled remote/persistent syslog storage for full crash data

Final syslog snippet before crash
Jul 30 02:32:48 NAS emhttpd: Unregistered Flash device blacklisted (EBLACKLISTED2)

Jul 30 02:32:48 NAS kernel: __vm_enough_memory: pid: 1501545, comm: udevadm, bytes: 100056763420672 not enough memory for the allocation

Jul 30 02:32:48 NAS kernel: udevadm[1501545]: segfault at 24fbc3ac1b70 ip 000014fbc3f32cc4 sp 00007ffc0b856510 error 6 in ld-linux-x86-64-2.41.so[ecc4,14fbc3f25000+2c000] likely on CPU 14 (core 30, socket 0)

Jul 30 02:32:48 NAS kernel: Code: 00 00 0f 1f 40 00 48 8b 31 8b 51 08 4c 01 fe 48 83 fa 26 74 0a 48 83 fa 08 0f 85 17 29 ff ff 48 8b 51 10 48 83 c1 18 4c 01 fa <48> 89 16 48 39 d9 72 d4 4d 8b 9a 08 02 00 00 4d 85 db 0f 84 64 0b

Jul 30 02:32:48 NAS kernel: __vm_enough_memory: pid: 1501546, comm: zfs, bytes: 99997216514048 not enough memory for the allocation

Jul 30 02:32:48 NAS kernel: zfs[1501546]: segfault at 250ca4ca6b70 ip 0000150ca5219cc4 sp 00007fff0f5decf0 error 6 in ld-linux-x86-64-2.41.so[ecc4,150ca520c000+2c000] likely on CPU 15 (core 31, socket 0)

Jul 30 02:32:48 NAS kernel: Code: 00 00 0f 1f 40 00 48 8b 31 8b 51 08 4c 01 fe 48 83 fa 26 74 0a 48 83 fa 08 0f 85 17 29 ff ff 48 8b 51 10 48 83 c1 18 4c 01 fa <48> 89 16 48 39 d9 72 d4 4d 8b 9a 08 02 00 00 4d 85 db 0f 84 64 0b

Jul 30 02:32:48 NAS kernel: __vm_enough_memory: pid: 1501548, comm: zfs, bytes: 100216445968384 not enough memory for the allocation

Jul 30 02:32:48 NAS kernel: zfs[1501548]: segfault at 24d7bbff0b70 ip 000014d7bc563cc4 sp 00007ffd31aaa790 error 6 in ld-linux-x86-64-2.41.so[ecc4,14d7bc556000+2c000] likely on CPU 12 (core 28, socket 0)

Jul 30 02:32:48 NAS kernel: Code: 00 00 0f 1f 40 00 48 8b 31 8b 51 08 4c 01 fe 48 83 fa 26 74 0a 48 83 fa 08 0f 85 17 29 ff ff 48 8b 51 10 48 83 c1 18 4c 01 fa <48> 89 16 48 39 d9 72 d4 4d 8b 9a 08 02 00 00 4d 85 db 0f 84 64 0b

Jul 30 02:32:48 NAS kernel: __vm_enough_memory: pid: 1501549, comm: zfs, bytes: 100758978768896 not enough memory for the allocation

Jul 30 02:32:48 NAS kernel: zfs[1501549]: segfault at 245aa9196b70 ip 0000145aa9709cc4 sp 00007ffe703e9450 error 6 in ld-linux-x86-64-2.41.so[ecc4,145aa96fc000+2c000] likely on CPU 7 (core 12, socket 0)

Jul 30 02:32:48 NAS kernel: Code: 00 00 0f 1f 40 00 48 8b 31 8b 51 08 4c 01 fe 48 83 fa 26 74 0a 48 83 fa 08 0f 85 17 29 ff ff 48 8b 51 10 48 83 c1 18 4c 01 fa <48> 89 16 48 39 d9 72 d4 4d 8b 9a 08 02 00 00 4d 85 db 0f 84 64 0b

Jul 30 02:32:49 NAS emhttpd: Starter key detected, GUID: 154B-00EE-071C-49B61FA3F495 FILE: /boot/config/Starter.key


Can someone help me interpret this log? I believe it might be an issue with corrupt ram or potentially a mobo BIOS sleep state? These were the only real issues I had left to think of.

syslog.txt

Edited July 30, 2025Jul 30 by ryanp
word choice

Memtest is showing bad RAM, leaving this thread up in case someone else has a similar issue in the future.