Before I make anything worse, I figured I’d seek out some help from the experts. A couple of weeks ago, I noticed that my NAS was inaccessible via web GUI, and I couldn’t mount the share via SMB. I could access it via SSH, and all of the apps were working, so I rebooted the system. After that, it failed to boot completely, getting stuck in a kernel panic state. I ended up having to reinstall TrueNAS. Fortunately all of my data was intact and I imported my data pools no problem. It ran fine for about a week and then yesterday, I think the same thing happened. The web GUI is inaccessible, the share won’t mount via SMB. I didn’t get around to enabling SSH on this install, so I can’t check that. But all of my apps are working fine, including a file sharing app, so I can confirm all of my data is intact. I’m afraid to reboot based on what happened last time, so I’m hoping there might be something else I can try.
Some relevant details: I’m running 25.10 Community, installed on a 128 GB NVMe SSD. I have two data pools, one on an SSD and another on mirrored 12 TB HDDs. My network isn’t anything fancy, I’ve got a 2Gbps fiber connection coming into the house and two ethernet cables connected to the computer, both 2.5Gbps ports, which I’ve combined in a bridge. I’ve assigned a static IP within the TruenNAS GUI, but because I’m worried that this may be network interface-related, I want to point out that I didn’t deactivate either of the individual ethernet interfaces in TrueNAS. Near as I can tell, you can’t do that while keeping the bridge active, but maybe I’m wrong. I can ping the IP just fine. If it matters, the system has 23 GB of RAM and has never had any overheating issues.
I’d run the system 24/7 for about a year before the first time this happened a couple of weeks ago, and never had any issues. I hooked up a monitor to see if I could learn anything before shutting down, and I see mostly the same message repeated: “systemd- journald [788]: Failed to send WATCHDOG=1 notification message: Transport endpoint is not connected” I wasn’t able to learn anything super helpful when searching this message. The only other thing on the display at the time was this: “br1: received packet on eno1 with own address as source address” followed by a MAC address and vlan:0.
Any help would be great. I don’t really want to go through a reinstall again if I can help it. I’m firmly in the “knows just enough to be dangerous” camp, so apologies if I have overlooked anything obvious here.
There’s not much to my network. I have an ATT gateway device setup in IP passthrough, with my router set as the passthrough fixed MAC address. The router itself is a mesh system, with a main router and two satellites. Both satellites have a wired backhaul courtesy of a MoCA setup. My NAS is connected to one of those satellites with two ethernet cables. In TrueNAS, I configured a bridge with the two ethernet interfaces and assigned it a static IP. I have that IP reserved on my router. No firewall in place. My DNS is set to 1.1.1.1 and 1.0.0.1. Um… I can’t think of anything else that might matter?
Oh, here’s something - I just checked my router’s app (it’s a TP-Link Deco, which is decent but has virtually no web interface, so it’s frustrating to use sometimes as you’re pretty much limited to a mobile app) and the reserved IP assigned to my server shows a different MAC address than the MAC address currently connected as a wired device to the satellite my server is connected to. (It’s the only wired device on that satellite.) And if I tap through to that device, there’s a little warning message that says “Random MAC address detected.” I don’t think I knew that was a thing that could happen. Maybe I should try and disable the ability to assign random MAC addresses within my router interface, and then re-reserve the same IP using whatever MAC address the NAS is currently showing? Does that sound right?
From a networking perspective, two NICs from a bridge connected to the same switch sounds like a recipe for trouble. The Random MAC warning might be TP-Link seeing the same IP from the two NICs.
Does your TP-Link support LACP or teaming or bonding or trunking (I think there might be a couple over names that get used)? I am not Linux/Unix networking expert but from my understanding, Bridged NICs act like a switch and bonded NICs act like a single NIC, assuming the other end supports LACP.
Look up Spanning Tree Protocol, you should find good info about the ugliness that happens when there is a loop in a switching topology. I personally have never seen a switch that would auto setup trunking, only Spanning Tree which turns off a port to break loops.
