Periodic NIC issues (?) with Protectli Vault, Intel i226-V

[OPNenthu]

Edit: kernel for 26.7 is up as well:

# opnsense-update -zkr 26.7-igc2

I'm happy to try this as I'm on 26.7 now but I'm not sure if my symptoms indicate me as a test candidate.  After re-reading @fornax's posts and #62 especially, I think you guys might be targeting a different set of issues than I have.  It's been hard to tell.

Do you have field complaints other than this thread?  Would you be open to setting up a "Call for testing" thread with a description of the specific problem(s) that the OPNsense team is targeting, a description of who is/isn't a test candidate, etc.?

[BrandyWine]

This is a good read, seems more like router tuning.
https://docs.opnsense.org/manual/interfaces_settings.html

What's the status of your fix?

[fornax]

First time for everything, I guess... yesterday I had an issue that was only resolved after bouncing the WAN interface instead of LAN like every other time. In a fit of frustration I reinstalled the OS today, so it's on a fresh 26.7. That should rule out OS corruption or weird config issues from me tinkering with things. I also installed the -igc2 kernel. Will continue to monitor.

[BrandyWine]

First time for everything, I guess... yesterday I had an issue that was only resolved after bouncing the WAN interface instead of LAN like every other time. In a fit of frustration I reinstalled the OS today

And you are running latest NVM for the 226's ?

[fornax]

And you are running latest NVM for the 226's ?

As of post #29, yes.

[BrandyWine]

So the WAN "died", and then you installed fresh copy of 26.7? That's not really a good troubleshooting tactic, because now the issue (if it still exists) is reset.

Is WAN router or LAN switch 2.5G capable?

Even on 15.1 (26.7), verify EEE settings and look for events.

ethtool --show-eee igc0
ethtool -i igc0
and
nohup sh -c "dmesg -aw | grep -i -E 'igc|eee|energy|lpi|sleep|wake'" > watch-energy.txt 2>&1 &

its wrapped nohup so if your ssh dies the capture keeps running, you come back in to look at the txt file, etc. Kill that pid when done.

[Patrick M. Hausen]

ethtool --show-eee igc0
ethtool -i igc0

Code Select Expand

root@opnsense:~ # ethtool
ethtool: Command not found.

OPNsense isn't Linux.

[fornax]

So the WAN "died", and then you installed fresh copy of 26.7? That's not really a good troubleshooting tactic, because now the issue (if it still exists) is reset.

I wanted to rule out corruption in the install. Also, as this was my first experience with OPNSense, I did a fair amount of poking around and playing with settings, so I wanted a fresh start to rule out any possibility of lingering config weirdness. And those are ruled out now, as I had another incident last night.

Is WAN router or LAN switch 2.5G capable?

WAN router yes, LAN switch no.

Even on 15.1 (26.7), verify EEE settings and look for events.

Code Select Expand

root@scutum:~ # sysctl -a | grep igc | grep eee
hw.igc.eee_setting: 1
dev.igc.1.eee_control: 1
dev.igc.0.eee_control: 1
dmesg and syslog have never shown any random events. Everything is associated with bootup, manual interface bounces, or other actions on my part (eg config changes).