Unraid 6.12.10 - 10 gig nic disconnects when docker enabled.

[reheat]

I purchased a 10 gig nic (Tp-link tx401) and have been having issues. 

 

Everything works well if docker is not enabled. Speed tests and file transfer are good and a solid connection.

 

When I enable docker the connection dies after about 1-2 minutes. The actual light goes out on the nic and it is unrecoverable until I reboot.

 

If I go back to using my onboard 2.5g lan connection everything is fine. 

 

I've experimented with different network and docker configs (I currently use the macvlan fix) without success. I've tried disabling the onboard lan but makes no difference. 

 

The strange thing (to me) is that with identical settings the onboard 2.5g lan works fine with or without docker enabled. However, the 10g nic only works with docker disabled and completely stops working when I enable it. I am not using both at the same time. 

 

Would anyone have any ideas? Thank you.

 

 

phantom-diagnostics-20240629-1141.zip

[JorgeB]

That's a strange one, and nothing is jumping out to me in the logs, do you now the approximated timecode it happened?

[reheat]

No, sorry, I'll go back and reproduce it and get a more accurate time and repost.

[reheat]

Here is an updated diag. The timing of the network going down is Jul  2 08:41:53

 

Once the link goes down I can no longer access the docker page (connected directly to the machine via the gui option). It just spins waiting but everything else is still available.

 

Thanks.

 

 

phantom-diagnostics-20240702-0847.zip

[JorgeB]

13 minutes ago, reheat said:

down is Jul  2 08:41:53

That's right after boot? Server booted at 08:40:50, do you confirm that?

[reheat]

Yes, that's correct. Right after a booting.

[JorgeB]

Don't see any errors logged, but eth1 doesn't have an IP set, how are you getting/setting the IP for that NIC, is it connected to the same router?

[reheat]

it isn't connected to anything. I've been just using one nic. I'd hoped to just swap the 2.5 onboard to the 10g. Just one connection.

I've noticed that this seems to be happening right before the loss of connection that doesn't appear in the logs for the other nic that works correctly:

Jul  2 08:41:51 phantom network: reload service: nginx
Jul  2 08:41:51 phantom nginx: 2024/07/02 08:41:51 [alert] 10475#10475: *89 open socket #24 left in connection 7
Jul  2 08:41:51 phantom nginx: 2024/07/02 08:41:51 [alert] 10475#10475: *90 open socket #25 left in connection 14
Jul  2 08:41:51 phantom nginx: 2024/07/02 08:41:51 [alert] 10475#10475: *97 open socket #31 left in connection 15
Jul  2 08:41:51 phantom nginx: 2024/07/02 08:41:51 [alert] 10475#10475: *147 open socket #18 left in connection 16
Jul  2 08:41:51 phantom nginx: 2024/07/02 08:41:51 [alert] 10475#10475: *100 open socket #32 left in connection 18
Jul  2 08:41:51 phantom nginx: 2024/07/02 08:41:51 [alert] 10475#10475: *91 open socket #26 left in connection 20
Jul  2 08:41:51 phantom nginx: 2024/07/02 08:41:51 [alert] 10475#10475: aborting
 

[JorgeB]

Are you setting the 10GbE NIC as eth0 when you use it?

[reheat]

Yes. I've also completely disabled the onboard hardware and just had the 10g set as eth0 in case there was some soft of conflict. Didn't make a difference unfortunately.

[JorgeB]

Sorry no more ideas, the symptoms suggest an issue with that specific NIC or driver.

 

 

[reheat]

Thanks for looking. I'll check into the driver.