HL15 running TrueNAS Scale 24.10.2.2. MB Supermicro X11SPH-nCTPF, bios 3.5.
I’ve had my HL15 for over a year with no issues (since Feb 2025). The other day there was a power outage (not the first) and it wouldn’t come back online. It boots to the console setup and I reconfigured eno1 with the same static ip. However, in a shell the eno1 link shows no-carrier. I’ve reseated the cable and even swapped cables and changed the switch port - no luck. I also did the hard power off thing (discharge capacitors, wait) but no change. No eno1 errors in the system log.
Thoughts? Looks like the current version of the bios is 4.7 - recommend an update?
Thanks - Tom
Although “not the first”, there might have been some sort of power spike over the network associated with the outage. I assume from the description you don’t have a UPS for the HL15 or the switch? Also, assuming you aren’t using optics. When you say you swapped cables, are you running DAC and you swapped the enire cable with the transceivers, or are you running an RJ45 transceiver and just swapped the RJ45 cable?
There are some commands you can try to see if there is damage to the port(s) or transceiver. Depending on the results, you might need to get a PCIe NIC replacement, or a cheapo one for testing to at least 100% rule out the switch.
Trigger the ASIC Built-In Self-Test"
ip link set eno1 up
sudo ethtool -t eno1 offline
Inspect the SFP+ I2C Bus Telemetry:
sudo ethtool -m eno1
Scrape the Low-Level Driver Statistics:
sudo ethtool -S eno1
Check Kernel Advanced Error Reporting:
sudo dmesg | grep -iE "i40e|aer|eno1"
You can try a BIOS upgrade, but that’s probably not likely to fix this, and might introduce a new unknown that might cloud what actually happened.
I’m not sure if this helps or not, but judging from posts here and elsewhere the Intel X722 chip has a bit of a reputation of dying kinda easily. I hope your board is ok, but it also would not surprise me if what seemed like a “routine” power outage actually took it out.
1 Like
Thanks for the help - afraid none of the tests showed anything really. I think you’re right - it’s the NIC itself. This is just the regular RJ45 NIC - not the SFP+. Everything else on the network is fine so I don’t think it’s the switch.
No unused PCIe NICs around so I’m in the market. Any recommendations on a good one for this board?
I don’t think motherboard compatibility is usually an issue. I think it is more around your software and physical network requirements. The easiest “drop in” replacement without asking questions would probably be a NIC based on the X710-T2L chip. That is the same silicon generation and uses the same drivers as the X722. If you wanted to save money, you could go X540 or X550–prior gen–but those run hotter. If you wanted the more recent chipset from Intel for 10 Gbe that would be the E610.
But you could use this as an opportunity to go SFP+ or 25 Gb if that was in your plans.
It’s still interesting that none of those commands turned up any errors. What happens if you;
Take a known-good Cat6 cable and plug one end into eno1 and the other end directly into eno2. (You are plugging the server into itself). Bring both interfaces up in the OS so they attempt to negotiate a link.
sudo ip link set eno1 up
sudo ip link set eno2 up
Do the green link lights on the RJ45 ports turn on or remain completely dark?
1 Like