Bonding stopped working after power outage

[chrisk2305]

Hi,

I have a problem with my Dual 10Gbe NIC and bonding since a power outage a few days ago. After cold booting I noticed that my server had no connectivity. Bond was brought up normally though. Checked the switch config and everything seemed fine. Cleared mac adress table on the switch and rebooted - but still no luck.

I haven't seen any errors in the log.

I could ping other devices in the same subnet though but was not able to reach the gateway. Did a traceroute and it took 3000ms! to reach the gateway. Then I disconnected one of the two fibre cables and voila internet, etc. was working again.

Do you guys have any idea what the problem could be?

Thanks in advance,
Christian

[szatox]

I suppose that bond is connected with at least 2 wires to a managed switch.
Is the same mode configured on both ends of the link? Mismatch at this point will only let it work by accident (so you should have _sort_of_ connectivity), but the packet loss that can occur in such scenario would make any smart protocol repeat at reduced rate, and then reduce rate and repeat again, and reduce rate and repeat....

Just a guess.
Providing some more details on your setup and pointing out devices that were affected by power outage could allow for another guess.
Also, do you often restart pieces of your equipment? Perhaps you hotfixed setup in runtime on some device and forgot to make this change permanent.

[chrisk2305]

Hi,

sorry I did not provide enough infomation. Yes the bond consists of two LC Cables with the appropriate SFP+ Modules. Has been working for 6 months without a problem.

Switch is a D-Link DGS-1510-28X and the NIC in the Server is Dual Port 10GBe with Broadcom Chipset (NetXtreme II driver). Bond is configured via netctl.

here the output of the bonding status with one nic (eth3) disconnected:

[bbgermany]

Hi,

have you checked your switch, whether it still has a valid bond/lacp/etherchannel/portchannel configuration on the ports where your server is attached to?

greets, bb
_________________
Desktop: Ryzen 5 5600G, 32GB, 2TB, RX7600
Notebook: Dell XPS 13 9370, 16GB, 1TB
Server #1: Ryzen 5 Pro 4650G, 64GB, 16.5TB
Server #2: Ryzen 4800H, 32GB, 22TB

[chrisk2305]

yes I checked the switch and everything is fine there. I just rebooted the server with only one cable attached (which worked) and had no connectivty. Just out of curiosity I attached the second cable (bond worked fine and enslaved eth3) but still no connectivity. Then I removed the second calbe again and voila connectivity was there.

here the dmesg output:

[chrisk2305]

Hi again,

I double checked the switch config and saw that the protocol was changed to static instead of lacp.

Thanks for you help guys...I was just blind