Ahh!

I've spent a few days troubleshooting a network issue issue and finally narrowed it down to unRAID. I'll probably include way more in this than necessary so accept this TL;DR:

• My ARP table was picking up a macvlan network on my unRAID server and causing IP conflicts and unicast flooding (killing my network). I believe I fixed it by doing the commands below. I've seen a lot of people talk about ipvlan/macvlan but not really all that much on flipping between the two (which I can't currently do as ipvlan is greyed out).

I setup my unRAID server back in the early v6 days and she's been a champ moving between 3 different iterations of server. I used to have a full UniFi setup which I think was smart enough to hide this issue. I got a pretty nice pfSense Netgate box which had way more juice than my UDM-P which when I added some more security cameras started to get rather sluggish. Almost immediately, the pfSense didn't like my unRAID setup. I had a DHCP reservation on the pfSense for my unRAID host IP and MAC address. However, when my switches booted up, they had a conflict because another MAC address picked up my unRAID IP. I blocked the device on the switches and thought nothing more of it.

A few weeks later, I notice my WiFi go down but not any wired devices. My backups only run for a few minutes every day, so it took me a bit to realize it was a traffic issue. Running iperf3 from my desktop to unRAID caused my access points to clog up and go offline. Spent days trying to figure out why my computer having a clear connection to unRAID was causing my network to shout at all active switch ports. Since my desktop is 2.5GbE, enough traffic was coming back to overload my 1GbE access points (and the wired connections had just enough Mbps to do their things). I finally figured it out! In pfSense, when I checked the ARP table, it reported a different MAC for my unRAID box. I deleted the entry, ran a ping to unRAID from the pfSense and it scooped up the correct MAC and now I can push 2.5Gb all day to unRAID and the rest of my devices are in harmony... for ~10 minutes when the ARP table updated and scooped up that incorrect MAC.

I did the following:

I ran docker network ls and 'eth0' came back as the only macvlan network (the rest of the entries are bridge or host).

I ran docker network inspect eth0 and it came back with my unRAID host address as the AuxillaryAddress and no containers associated with it.

I ran ip addr show | grep "02:23:5a:74:98:ad which returned "link/ether 02:23:5a:74:98:ad brd ff:ff:ff:ff:ff:ff"

I ran docker network rm eth0

I'm posting this for a few reasons.

1. I have little hair left; I was ripping it out trying to figure this issue out. The fact that I did had me doing a very nauseating happy dance!

2. I'm an idiot, but I couldn't find similar sounding issues on forums (bear in mind, unRAID itself wasn't even on my radar as a suspect for a lot of troubleshooting). This post hopefully will help someone down the line. Also going to be updating my STH and Reddit posts to reflect the solution.

3. Is what I did the accurate way to close this saga, or should I look into what caused it to happen to begin with?

unRAID v7.2.3

Edited April 15Apr 15 by TheSystemAdmin
Correcting typos.

May have posted too soon... that MAC came back into my ARP table. The only other item I'm running is a Home Assistant VPN which has a MAC Address given to it.

7 hours ago, TheSystemAdmin said:

which I can't currently do as ipvlan is greyed out

This means bridging is disabled; you may want to retest with bridging enabled and ipvlan.