Hi,

I've been troubleshooting this for a week and have finally found what's causing it, I just don't know how to fix it.

 

I recently had my unRaid server die.  Probable motherboard issue.  So I bought new motherboard and new CPU (old CPU was 6600K).  New CPU is 13900F.

 

Some issues persist, as I wasn't planning on upgrading and so never got to undo some things before upgrading.  One thing was the old system had two NICs and I had them bonded in a failover.  The new system seems to be operating fine, but upon boot up, it seems to hang for a bit and then mentions network bond failed.  I don't know if this is causing the problems or not, but thought I'd mention it.  There's nothing to change in the network settings, but there is some config left over from the first machine.

The problem comes in when I start up Docker.  As long as Docker isn't running, everything on the network is fine.  When I start it up however, I have two MAC addresses getting the same IP.  The MAC address of the NIC has been entered into my DHCP server so unRaid always gets the same IP.  But there is another MAC address getting the same IP.  What ends up happening is something in docker fights my DHCP server, and then suddenly my entire network goes down.  Every device gets a self-assigned IP (169.254.x.x) for a few minutes until the DHCP server gets control again, then the IP addresses get assigned like their supposed to, and I get my network back.  Then it happens again a few minutes later.  The only way I have found to stop it is to shut Docker down on my unRaid server.

 

I don't know if that second MAC address getting the same IP is what is causing this, but I have tried to block that other MAC on the network.  When I do, unRaid/Docker just changes the MAC and the same behavior starts again.  I finally found what that MAC address belongs to today.

 

shim-eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 10.0.0.10  netmask 255.255.255.255  broadcast 0.0.0.0
        inet6 fe80::582b:1cff:feb4:5806  prefixlen 64  scopeid 0x20<link>
        ether 5a:2b:1c:b4:58:06  txqueuelen 1000  (Ethernet)
        RX packets 21069  bytes 4794289 (4.5 MiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 188380  bytes 20486189 (19.5 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

 

This interface doesn't exist unless Docker is running.

 

And this is what shows up in my DHCP logs when the issue happens:

Date				Facility	Severity	Process	PID	Line
2023-05-12T10:57:49-04:00	23	Error	dhcpd	67826	dhcp.c:4164: Failed to send 300 byte long packet over igc1 interface.	
2023-05-12T10:57:49-04:00	23	Error	dhcpd	67826	send_packet: No buffer space available	
2023-05-12T10:57:46-04:00	23	Error	dhcpd	67826	dhcp.c:4164: Failed to send 300 byte long packet over igc1 interface.	
2023-05-12T10:57:46-04:00	23	Error	dhcpd	67826	send_packet: No buffer space available	
2023-05-12T10:57:46-04:00	23	Error	dhcpd	67826	dhcp.c:4164: Failed to send 300 byte long packet over igc1 interface.	
2023-05-12T10:57:46-04:00	23	Error	dhcpd	67826	send_packet: No buffer space available	
2023-05-12T10:57:44-04:00	23	Error	dhcpd	67826	dhcp.c:4164: Failed to send 300 byte long packet over igc1 interface.	
2023-05-12T10:57:44-04:00	23	Error	dhcpd	67826	send_packet: No buffer space available	
2023-05-12T10:57:38-04:00	23	Error	dhcpd	67826	dhcp.c:4164: Failed to send 300 byte long packet over igc1 interface.	
2023-05-12T10:57:38-04:00	23	Error	dhcpd	67826	send_packet: No buffer space available	
2023-05-12T10:57:24-04:00	23	Error	dhcpd	67826	dhcp.c:4164: Failed to send 300 byte long packet over igc1 interface.	
2023-05-12T10:57:24-04:00	23	Error	dhcpd	67826	send_packet: No buffer space available	
2023-05-12T10:57:24-04:00	23	Error	dhcpd	67826	dhcp.c:4164: Failed to send 300 byte long packet over igc1 interface.	
2023-05-12T10:57:24-04:00	23	Error	dhcpd	67826	send_packet: No buffer space available	
2023-05-12T10:57:22-04:00	23	Error	dhcpd	67826	dhcp.c:4164: Failed to send 300 byte long packet over igc1 interface.	
2023-05-12T10:57:22-04:00	23	Error	dhcpd	67826	send_packet: No buffer space available	
2023-05-12T10:57:21-04:00	23	Error	dhcpd	67826	dhcp.c:4164: Failed to send 300 byte long packet over igc1 interface.	
2023-05-12T10:57:21-04:00	23	Error	dhcpd	67826	send_packet: No buffer space available

 

Does anyone have any idea what I need to do to stop this?  I've included the server diagnostics if it's needed.

 

Thanks,
Scott

 

server-diagnostics-20230513-1627.zip

You might want to delete/rename /boot/config/network.cfg and reboot so that unraid creates a new one with defaults for this machine's config, then reconfigure what's needed.

Thanks.  I should have thought about that to begin with.  That took care of the old bonded interface, but not the network outages.

 

Before I start docker, there are only 2 entries in the routing table:

default            10.0.0.10 via br0

10.0.0.10/24     br0

 

When docker starts, I get the routing table attached.  Using the 'docker network ls' and 'docker network inspect', the entries for 172.17.0.0/16 and 172.20.0.0/16 appear.  I don't know where the shim-br0 entries are coming from.  They're obviously related to docker, but don't show up with the 'docker network' commands, and my theory is these are the problems.  Unable to delete them. Only way to get rid of them is to disable Docker again.

38 minutes ago, Aquenon said:

I don't know where the shim-br0 entries are coming from.

 

Thanks Hoopster.  Not sure why I couldn’t find that.  I unchecked this option and I haven’t seen this error again since so I’ll mark your post as a solution.  Though I had this exact same setup on the old hardware, and it never caused problems on the network.