TrueNAS Mini X Losing Network Connection

[xoomdust]

I have a new as of late last year TrueNAS Mini X running TrueNAS-12.0-U8.1. Every once in a while, seemingly during Veeam backups to the SMB share, it loses network connection, and I have to reboot it from the IPMI to restore the network connection. It will usually work for a month or two after that, then disconnect again.

However, today, I had to reboot it, and ran a backup to test it, and it disconnected again right away. I have tried different LAN ports on the device, different switches, switch ports, and changed to brand new cables. This only started happening once I moved it to a static IP, as I had it on DHCP with a reservation before we moved it to an isolated VLAN with DHCP off.

Can anyone provide some insight as to what's going on? I never had this problem with custom-built FreeNAS boxes.

 

[Nick2253]

If it's only a year old, then I'd guess you'd have warranty. I'd contact iXsystems and use that support directly.

If it's not under warranty, then, from a cost-benefit perspective, I would just add a separate NIC PCIe card and be done with it. Troubleshooting these problems will easily take a few hours at least, and a new Intel quad-port PCIe NIC can be had for under $100 (and 1 and 2-port cards are even cheaper). I know it's not the "best" option, but I can't imagine that the NIC will cost more than your time.

If you really do want to do troubleshooting, then it's possible that something is going wrong in the hardware (also why I just suggest getting a new NIC). Those symptoms sound like potential overheating: lots of data starts to move through the network adapter, adapter chip heats up, ends up in some funky state, and then it requires a reboot to clear. If you have physical access, I would open it up, and blow out all the dust, and make sure that there are no signs of damage near the network ports.

To explicitly rule out an overheating problem, I'd put a fan directly blowing over the motherboard chips. If you can repeat the problem with all that airflow, then I'd call my hardware theory successfully debunked.