So tonight I was mucking about trying to figure out why the apps tab on my unraid server wouldn't load after a reboot. In the process of diagnostic, I unplugged the ethernet cable and plugged it back in. After that, I was no longer able to access my server in any capacity. No webgui, no access to my array. I tried deleting the network.cfg to generate a fresh one and nothing. The IP it shows on startup is 169.254.208.234, but when I try to connect to that via HTTP://169.254.208.234, I get HTTP://169.254.208.234 is unreachable. I'm new to unraid and an event like this is leaving me very lost. I've attached the diagnostics file if anyone can help me. Still don't know why the apps page won't load after reboot.

 

 

tower-diagnostics-20241225-1938.zip

the "169..." is a clear sign that your UNRAID cannot reach a dhcp server to get a "real" address.

So check out if it is connected to your router (or whatever box in your LAN runs the dhcp server).

 

The apps page not loading is just a result of the missing dhcp server. Dont worry about this for now, once the address (and the routing) has been fixed, the apps will automatically come back.

 

Wowsers! You have put 4 Nics into 1 bond! one of them not even exisisting in the box anymore! Delete this bond instantly! you dont need it and it will only give you grief.

 

Delete the bond, plug the cable into the Realtek card and reboot. This should bring you back into buisiness. Then you might explain what you want to have done with the rest of the cards and we can debate how to get there.

 

I guess you want to run UNRAID on one of the two Intel 10G cards, but this needs to be set up a bit more tricky.

Report back when you are running well again on 1G...

Edited December 26, 20241 yr by MAM59

Bond deleted, plugged into realtek card, still nothing. Rebooting seems to reinitiate the bond, so I'm guessing I need to disable bonding altogether? I can't find how to do that online without using the GUI, which I cannot access. The "two" intel cards are actually one card with two ports. I would like to use one of them for my unraid. Assuming I need to disable bonding, how can I do that without GUI access?

Yeah I know that this is one card with 2 ports.

But we will deal with it later.

 

You cannot delete /boot/network.cfg from a running unraid, it will be recreated at shutdown.

So, to get rid of it, shut down unraid, put the stick into a different computer (without booting it from the stick of course!) and delete it from there.

 

This should get you rid of the evil bond line

BONDNICS[0]="eth0 eth1 eth2 eth3"

and the even more evil

BRNICS[0]="bond0" (should be "eth0"!)

 

Edited December 26, 20241 yr by MAM59

When I delete the network.cfg from another computer, it regenerates with the same settings, bonding included.

7 minutes ago, cheesysenator said:

When I delete the network.cfg from another computer, it regenerates with the same settings, bonding included.

Hardly believable, sorry.

should create something like:

 

IFNAME[0]="br0"
BRNAME[0]="br0"
BRSTP[0]="no"
BRFD[0]="0"
DESCRIPTION[0]="Lokales LAN"
BRNICS[0]="eth0"
PROTOCOL[0]="ipv4+ipv6"
USE_DHCP[0]="no"
IPADDR[0]="192.168.0.4"
NETMASK[0]="255.255.255.0"
GATEWAY[0]="192.168.0.253"
METRIC[0]="1"
DNS_SERVER1="192.168.0.8"
DNS_SERVER2="192.168.0.9"
USE_DHCP6[0]="no"

 

(if you set USE_DHCP[0]="yes", you can leave out IPADDR, NETMASK, GATEWAY and DNS_SERVER?)

Try to edit it manually on a different computer.

 

The bond by itself is not a problem, the problem is that you are getting an APIPA IP address, suggesting the server is not correctly connected to the DHCP server, typically your router, or there are issues with it.

 

I see that there are two NICs with a link, intel and Realtek, which one do you plan to use as your main NIC? Disconnect the cable from the other one for now.

On 12/26/2024 at 10:07 AM, MAM59 said:

Hardly believable, sorry.

should create something like:

 

IFNAME[0]="br0"
BRNAME[0]="br0"
BRSTP[0]="no"
BRFD[0]="0"
DESCRIPTION[0]="Lokales LAN"
BRNICS[0]="eth0"
PROTOCOL[0]="ipv4+ipv6"
USE_DHCP[0]="no"
IPADDR[0]="192.168.0.4"
NETMASK[0]="255.255.255.0"
GATEWAY[0]="192.168.0.253"
METRIC[0]="1"
DNS_SERVER1="192.168.0.8"
DNS_SERVER2="192.168.0.9"
USE_DHCP6[0]="no"

 

(if you set USE_DHCP[0]="yes", you can leave out IPADDR, NETMASK, GATEWAY and DNS_SERVER?)

Try to edit it manually on a different computer.

 

Definitely does regenerate config with bonding enabled, although I don't know why that would be the problem. If I set USE_DHCP to 'yes', then it fails to contact the DHCP server for an IP, staying in the 169 range, and is unreachable. If I set it to 'no', I specify the IP, and I can see the IP in my network from my desktop, but it is unreachable. All pings will timeout. I have tried it now with and without bonding, by editing the file to remove the bonding lines. Neither way has worked.

On 12/26/2024 at 12:51 PM, JorgeB said:

The bond by itself is not a problem, the problem is that you are getting an APIPA IP address, suggesting the server is not correctly connected to the DHCP server, typically your router, or there are issues with it.

 

I see that there are two NICs with a link, intel and Realtek, which one do you plan to use as your main NIC? Disconnect the cable from the other one for now.

Planning to use the intel NIC. The other one doesn't even have a cable connected. The connection to the router should be fine, there is another computer connected to the same switch that is functioning fine. Tried different ports and cables to no avail.

As I already said above and long ago: use the 1G card for now!

I know this is not what you want finally, but you need to get into business again first before experiments can happen.

Leave the Intel unconnected, put the cable into the 1G and reboot. The box should be reachable again and we can discuss further actions needed for switching to the intel card then.

 

9 minutes ago, MAM59 said:

As I already said above and long ago: use the 1G card for now!

I know this is not what you want finally, but you need to get into business again first before experiments can happen.

Leave the Intel unconnected, put the cable into the 1G and reboot. The box should be reachable again and we can discuss further actions needed for switching to the intel card then.

 

Tried that several times during this process. It did not work.

ok, this means you do not have an unraid problem, but there is something wrong with cable/switch/vlan (aka: ELSEWHERE)

 

If you plug a card (if all dont work it does not matter which one, but I would still use the 1G  because there are fewer points of failure) into the switch, do you get a link? (LEDs turn on on both ends?) (both! hopefully there are LEDs on both ends, just one side does not mean a thing because each side only checks half of the cable)

 

18 minutes ago, MAM59 said:

ok, this means you do not have an unraid problem, but there is something wrong with cable/switch/vlan (aka: ELSEWHERE)

 

If you plug a card (if all dont work it does not matter which one, but I would still use the 1G  because there are fewer points of failure) into the switch, do you get a link? (LEDs turn on on both ends?) (both! hopefully there are LEDs on both ends, just one side does not mean a thing because each side only checks half of the cable)

 

Yes, leds on both ends light up and blink. On motherboard eth port and eth switch

Edited December 28, 20241 yr by cheesysenator
added detail

aha, and you are sure that there is no VLAN configured or a firewall setting somewhere that prevents communication between the ports???

 

12 minutes ago, MAM59 said:

aha, and you are sure that there is no VLAN configured or a firewall setting somewhere that prevents communication between the ports???

 

Shouldn't be anything. I know I haven't configured anything like that. The unraid server is plugged into an unmanaged switch which is running other devices without issue.

hmmm, I'm afraid, I am running out of options, sorry.

Next I would try would be to take out the intel card completely and see if the 1G resumes work again. Maybe a PCIe problem or something.

But, as you can see, this is only Try&Error. So I am afraid, I have to leave you here.

 

3 hours ago, cheesysenator said:

Planning to use the intel NIC. The other one doesn't even have a cable connected.

Yep, sorry, misread the diags the first time, is the 10GbE NIC connected to the same DHCP server? Presumably your router?

 

I would also recommend connecting just the gigabit NIC for now, unplug the other one and post new diags, if it also gets an APIPA address, the issue is most likely the DHCP server.

maybe just one faint and unlikely idea:

 

many of these cheap unmanaged switches have a switch (lol!) called "VLAN" or something. With this one turned on, Traffic between Ports is forbidden, all Ports can only "talk" to one dedicated Port.

Of course for normal operations this needs to be turned OFF.

 

So maybe check for hidden switches (or look into the manual).

(And yeah, I also dont know why they now built it in. Makes very very few sense to me)

 

I have completely removed the intel nic and plugged into the 1 gig on-board card. It is still not generating a proper IP address. I have attached a new diagnostics file. There is also not a vlan switch on my switch.

tower-diagnostics-20241228-2127.zip

Edited December 29, 20241 yr by cheesysenator

Sorry again, but its (almost) the same as before:

* card found: OK

* card initiliased: OK

* link established: OK

* asking for DHCP Server: FAILURE

 

Still your UNRAID cannot reach any dhcp server in your LAN. The reason for this must be outside the unraid box.

Where is/should your DHCP Server? in your Router??? on a different machine?

 

39 minutes ago, MAM59 said:

Sorry again, but its (almost) the same as before:

* card found: OK

* card initiliased: OK

* link established: OK

* asking for DHCP Server: FAILURE

 

Still your UNRAID cannot reach any dhcp server in your LAN. The reason for this must be outside the unraid box.

Where is/should your DHCP Server? in your Router??? on a different machine?

 

I suppose the DHCP server is run on the fiber box provided by my ISP. I have tried restarting that a number of times to no avail. Could it be some weirdness since I'm running straight back there rather than through a router? My PC works fine with the same configuration... 

9 minutes ago, cheesysenator said:

My PC works fine with the same configuration... 

Pls go to your PC, open a cmd window and type in "ipconfig /all" (assuming it is running a kind of windows).

Post the output here.

 

Update: also the output of "netstat -nr" could be valuable to take a look at. add it here too pls

 

Edited December 29, 20241 yr by MAM59

On 12/29/2024 at 1:51 AM, MAM59 said:

Pls go to your PC, open a cmd window and type in "ipconfig /all" (assuming it is running a kind of windows).

Post the output here.

 

Update: also the output of "netstat -nr" could be valuable to take a look at. add it here too pls

 

Here they are.

 

 

Edited December 30, 20241 yr by cheesysenator

AAARGGGHHH!!!

You are using a Wireguard tunnel and your DHCP Server is "at the other end" of the tunnel.

Of course UNRAID cannot reach it (and never will be able to do it)

 

No idea what this is good for and also no idea how to overcome it, sorry.

 

 

5 minutes ago, MAM59 said:

AAARGGGHHH!!!

You are using a Wireguard tunnel and your DHCP Server is "at the other end" of the tunnel.

Of course UNRAID cannot reach it (and never will be able to do it)

 

No idea what this is good for and also no idea how to overcome it, sorry.

 

 

What does the wireguard tunnel on my PC have to do with unraid being unable to reach the DHCP server?