Is there a way to recover from "Detected Tx Unit Hang" network errors?

Question

After waking from suspend, my Ubuntu 21.04 system had network issues. I'd guess about 50% of IP sockets failed. This included failures of both ssh and ping. By "fail" I mean that a connection attempt blocks, and then eventually times-out. (Aside: maybe ping doesn't use sockets?) I believe all programs that attempted network access encountered the same failure rate.

I've been suspending the system every night for several weeks. Today was the first time I had any problems after resuming.

I rebooted the system, which seems to have solved the problem (at least until it happens again).

After rebooting, I discovered various messages in syslog.

I see the below two lines on every suspend:

Sep  4 09:00:18 hostname kernel: [896165.908582] igc 0000:b0:00.0: no suspend buffer for PTM
Sep  4 09:00:18 hostname kernel: [896165.908589] igc 0000:af:00.0: no suspend buffer for PTM

(Aside: While the above lines refer to events that happened prior to the suspend, I believe the log entries were only created after the system resumed. Consequently, the events actually happened hours before the indicated times.)

And two identical lines on every resume:

Sep  4 09:00:18 hostname kernel: [896166.257107] igc 0000:af:00.0: no suspend buffer for PTM
Sep  4 09:00:18 hostname kernel: [896166.257177] igc 0000:b0:00.0: no suspend buffer for PTM

(Aside: Maybe there are two lines because the system has two built-in NICs?)

And here are the log entries that only occurred on my most recent resume, which is when the network problems started:

Sep  4 09:00:19 hostname kernel: [896168.214463] igc 0000:b0:00.0 enp176s0: Register Dump
Sep  4 09:00:19 hostname kernel: [896168.214467] igc 0000:b0:00.0 enp176s0: Register Name   Value
Sep  4 09:00:19 hostname kernel: [896168.214478] igc 0000:b0:00.0 enp176s0: CTRL            081c0641
Sep  4 09:00:19 hostname kernel: [896168.214481] igc 0000:b0:00.0 enp176s0: STATUS          40380401
Sep  4 09:00:19 hostname kernel: [896168.214483] igc 0000:b0:00.0 enp176s0: CTRL_EXT        10000040
Sep  4 09:00:19 hostname kernel: [896168.214486] igc 0000:b0:00.0 enp176s0: MDIC            18017949
Sep  4 09:00:19 hostname kernel: [896168.214488] igc 0000:b0:00.0 enp176s0: ICR             00000001
Sep  4 09:00:19 hostname kernel: [896168.214490] igc 0000:b0:00.0 enp176s0: RCTL            04408022
Sep  4 09:00:19 hostname kernel: [896168.214497] igc 0000:b0:00.0 enp176s0: RDLEN[0-3]      00001000 00001000 00001000 00001000
Sep  4 09:00:19 hostname kernel: [896168.214504] igc 0000:b0:00.0 enp176s0: RDH[0-3]        00000000 00000000 00000000 00000000
Sep  4 09:00:19 hostname kernel: [896168.214511] igc 0000:b0:00.0 enp176s0: RDT[0-3]        000000ff 000000ff 000000ff 000000ff
Sep  4 09:00:19 hostname kernel: [896168.214518] igc 0000:b0:00.0 enp176s0: RXDCTL[0-3]     02040808 02040808 02040808 02040808
Sep  4 09:00:19 hostname kernel: [896168.214525] igc 0000:b0:00.0 enp176s0: RDBAL[0-3]      2e35d000 51576000 337b2000 29a5b000
Sep  4 09:00:19 hostname kernel: [896168.214532] igc 0000:b0:00.0 enp176s0: RDBAH[0-3]      00000001 00000001 00000002 00000001
Sep  4 09:00:19 hostname kernel: [896168.214534] igc 0000:b0:00.0 enp176s0: TCTL            a50400fa
Sep  4 09:00:19 hostname kernel: [896168.214541] igc 0000:b0:00.0 enp176s0: TDBAL[0-3]      2e35d000 51576000 337b2000 29a5b000
Sep  4 09:00:19 hostname kernel: [896168.214548] igc 0000:b0:00.0 enp176s0: TDBAH[0-3]      00000001 00000002 00000001 00000001
Sep  4 09:00:19 hostname kernel: [896168.214555] igc 0000:b0:00.0 enp176s0: TDLEN[0-3]      00001000 00001000 00001000 00001000
Sep  4 09:00:19 hostname kernel: [896168.214562] igc 0000:b0:00.0 enp176s0: TDH[0-3]        00000007 00000009 00000004 00000000
Sep  4 09:00:19 hostname kernel: [896168.214568] igc 0000:b0:00.0 enp176s0: TDT[0-3]        0000000d 00000011 00000004 00000001
Sep  4 09:00:19 hostname kernel: [896168.214575] igc 0000:b0:00.0 enp176s0: TXDCTL[0-3]     02100108 02100108 02100108 02100108
Sep  4 09:00:19 hostname kernel: [896168.214576] igc 0000:b0:00.0 enp176s0: Reset adapter

And then there were many (over 19,000) entries like the below:

Sep  4 09:01:22 hostname kernel: [896231.189109] igc 0000:b0:00.0 enp176s0: Detected Tx Unit Hang
Sep  4 09:01:22 hostname kernel: [896231.189109]   Tx Queue             <1>
Sep  4 09:01:22 hostname kernel: [896231.189109]   TDH                  <9>
Sep  4 09:01:22 hostname kernel: [896231.189109]   TDT                  <9>
Sep  4 09:01:22 hostname kernel: [896231.189109]   next_to_use          <9>
Sep  4 09:01:22 hostname kernel: [896231.189109]   next_to_clean        <9>
Sep  4 09:01:22 hostname kernel: [896231.189109] buffer_info[next_to_clean]
Sep  4 09:01:22 hostname kernel: [896231.189109]   time_stamp           <10d596f4c>
Sep  4 09:01:22 hostname kernel: [896231.189109]   next_to_watch        <000000002b25478e>
Sep  4 09:01:22 hostname kernel: [896231.189109]   jiffies              <10d59ad90>
Sep  4 09:01:22 hostname kernel: [896231.189109]   desc.status          <0>

My question is, other than a full system reboot, is there some way I can manually reset/reboot the network interface from the command line?

The system is a desktop workstation, so I have keyboard access.

After rebooting, I realized that maybe I should try re-suspending and re-waking the system. But I only thought of that after doing the full reboot.

Answer 1

I encountered the network error again after waking from sleep. This time, I did not reboot, but instead tried the below (fairly obvious) solution:

sudo  ifconfig  enp176s0  down
sudo  systemctl  stop   systemd-networkd
sudo  systemctl  start  systemd-networkd

And my network is working again!

If you encounter the problem:

Replace enp176s0 with the name of your network interface.

Your network manager may be something other than systemd-networkd.

If you use a graphical desktop environment, then instead of the above, you may be able to simply disable and then re-enable networking via a menu on your desktop toolbar.