Network configuration

[dnLL]

I have a very small issue and have no idea what I changed to create the issue (it used to work properly).

 

I have multiple computers including my server on the same network hooked up to the same router. They all get a DHCP address from the router. From any computer on the network, I can do a local ping to any other computer or to the server. I can also ping the VMs, no issues. However, from the server and from the VM, same result: I can NOT ping any of the other computer on the network (I can ping the server/VMs from the server/VMs just fine however). I could understand the difficulties coming from the VMs but from the server itself I don't get it, plus it used to work in the past. 

 

I'm not sure what that means but here are my network settings:

 

 

 

As for my router, I flashed the FW to factory default recently and it didn't change anything... now, it has a DHCP reservation for my server but that's about it.

 

Why can't my server ping the other devices on the network?

 

It you think it's unrelated completely to my unRAID configuration, let me know... but I have no issue with any other device on the network.

Edited May 25, 2018 by dnLL

[Tybio]

Interesting, if you fire up a console, do you see any arp entries after you try to pinging outbound?  I just ran a ping from my server to my laptop....

 

For example (Removed the Docker lines and other stuff, my laptop is .7 and gateway is .1):

 

root@Storage:~# arp -a
Tybio-Laptop.fios-router.home (192.168.1.7) at 78:4f:43:55:19:ad [ether] on br0
FIOS_Quantum_Gateway.fios-router.home (192.168.1.1) at 48:5d:36:71:6f:6f [ether] on br0
root@Storage:~#

 

[dnLL]

5 hours ago, Tybio said:

Interesting, if you fire up a console, do you see any arp entries after you try to pinging outbound? 

 

There we go.

root@server:~# arp -a
DD-WRT (192.168.1.1) at 14:91:82:c4:6a:eb [ether] on br0
Chromecast (192.168.1.230) at 88:3d:24:22:13:54 [ether] on br0
ipmi (192.168.1.100) at d0:50:99:e1:db:ea [ether] on br0

root@server:~# ping 192.168.1.100
PING 192.168.1.100 (192.168.1.100) 56(84) bytes of data.
--- 192.168.1.100 ping statistics ---
9 packets transmitted, 0 received, 100% packet loss, time 8188ms

root@server:~# ping 192.168.1.230
PING 192.168.1.230 (192.168.1.230) 56(84) bytes of data.
64 bytes from 192.168.1.230: icmp_seq=1 ttl=64 time=3.76 ms
64 bytes from 192.168.1.230: icmp_seq=2 ttl=64 time=5.03 ms
64 bytes from 192.168.1.230: icmp_seq=3 ttl=64 time=5.38 ms
64 bytes from 192.168.1.230: icmp_seq=4 ttl=64 time=4.54 ms
--- 192.168.1.230 ping statistics ---
4 packets transmitted, 4 received, 0% packet loss, time 3005ms
rtt min/avg/max/mdev = 3.762/4.682/5.387/0.613 ms

This is weird to be honest. From my desktop, no problem...

 

 

And it used to work from the Linux servers before. I didn't update the IPMI interface on my ASRock mobo (and don't see how that would be relevant anyway). Even the HTTP (ipmi) webgui used to be accessible from the server and its VMs, now it's not anymore...

[John_M]

Post your diagnostics zip.

[bonienl]

14 hours ago, dnLL said:

Why can't my server ping the other devices on the network?

 

Check if any firewall on these devices is interfering.

[dnLL]

On 5/26/2018 at 6:50 AM, John_M said:

Post your diagnostics zip.

 

 

 

On 5/26/2018 at 6:54 AM, bonienl said:

 

Check if any firewall on these devices is interfering.

 

It's a simple IPMI port, I highly doubt there are any firewall and I'm not sure why the firewall would accept pings from all devices but the server and its VMs. But I'll do some more testing for sure.

Edited May 28, 2018 by dnLL

[John_M]

@dnLL Would you mind trying something for me, please? Go to Settings -> Network Settings in the GUI, and under the heading Interface eth0 set Enable bonding to Yes. You'll need to stop the Docker and VM services first to be able to do this - probably the easiest way is simply to stop the array. Leave the bonding mode set to active-backup and leave eth0 as the sole member of bond0, like this:

 

 

Now click the Apply button, then restart the array. Now do your ping test again and report back.

 

I notice your motherboard has two main NICs (plus the IPMI one). At the moment you're only using eth0 (the one using the i210 chip), and eth1 is unplugged. Have you decided how you want to use this second port?

 

EDIT:

I should add that my reason for suggesting that you enable bonding is because that appears to be the default unRAID setting, even if only a single NIC is present. One of my servers has only one NIC and has bridging enabled, but it also has bonding enabled (per the screenshot) and I have no problem pinging.

 

Edited May 26, 2018 by John_M

[dnLL]

38 minutes ago, John_M said:

@dnLL Would you mind trying something for me, please? Go to Settings -> Network Settings in the GUI, and under the heading Interface eth0 set Enable bonding to Yes. You'll need to stop the Docker and VM services first to be able to do this - probably the easiest way is simply to stop the array. Leave the bonding mode set to active-backup and leave eth0 as the sole member of bond0, like this:

 

 

Now click the Apply button, then restart the array. Now do your ping test again and report back.

 

I notice your motherboard has two main NICs (plus the IPMI one). At the moment you're only using eth0 (the one using the i210 chip), and eth1 is unplugged. Have you decided how you want to use this second port?

 

EDIT:

I should add that my reason for suggesting that you enable bonding is because that appears to be the default unRAID setting, even if only a single NIC is present. One of my servers has only one NIC and has bridging enabled, but it also has bonding enabled (per the screenshot) and I have no problem pinging.

 

 

I remember playing with these settings so that might be the issue. I changed it and restarted the array but I noticed my SSH connection to the server wasn't interrupted or anything and it's still not working. I'm wondering if I need to reboot completely (I'm rebooting it right now, I'll check again once I'm back home later today).

 

The other NIC isn't used right now and I'm not sure I'm getting any benefit from using it... I have an ASRock Rack E3C236D2I motherboard. I get I could get 2 Gbit ethernet instead of 1 but... I'm limited by the speed of my HDDs at around 110 MB/s anyway (until I add SSD cache) and even if 2 different HDDs are being used, I rarely (I'd say never but I guess there are some very corner case scenarios) transfer files at full gigabit speed multiple times simultaneously.

Edited May 26, 2018 by dnLL

[dnLL]

 

It doesn't work after a reboot either... ?

[John_M]

I was going to suggest that you reboot after making the change, but you did anyway. As that didn't fix it I don't know what to suggest next.

 

But does it actually matter? If your server otherwise works normally you could just call it a strange quirk and live with it. If you really want to persevere you could trash your network settings and set them up from the beginning again. If you're interested, they are the following two files:

• /boot/config/network.cfg
• /boot/config/network-rules.cfg

as seen from the command prompt. However, I don't see anything wrong with them. Just delete them and reboot then set up networking again, though the defaults should be pretty close. The default should have bonding enabled and both NICs as members of the bond, so it won't matter which port you plug your network cable into. That might be the best use of the second port at the moment - a redundant spare. The bond (bond0) gets it's MAC address from eth0 so if you've set up your router's DHCP server to allocate a specific IP address that will still work.

[dnLL]

3 hours ago, John_M said:

But does it actually matter? If your server otherwise works normally you could just call it a strange quirk and live with it.

 

Actually it matters for my Check_MK server since it can't ping anything local outside the server currently (well it can apparently ping the chromecast which isn't exactly very interesting).

 

I'll try to reset the settings a little bit later and if that doesn't work... I will find a workaround I guess. I'm still quite intrigued since it was working when I added my ipmi to check_mk and I even added my desktop for testing purposes and it can't ping it anymore which is weird.

Edited May 26, 2018 by dnLL

[Geck0]

Hi,

I'm a first time poster. I'm struggling to get my network configuration correct. So, the default setting of bonding starts up okay. What I've been trying to do is turn off bonding, and have bridging enabled for both ports.

 

Currently, my server's two gigabit ports are plugged into a Netgear wireless extender connecting to my router. This is temporary until I can find time to wire in more network ports at home.

What I'm attempting to do is have dockers running for typical services. Example:

• Nextcloud
• Calibre
• Sonarr
• Kodi

Ideally these would be running on the first ethernet bridge. I would then like to have another dedicated ethernet port bridged with a static ip address that will act as reverse proxy, multi factor authentication, etc. Would this be better served through a VM rather than a docker,  I wonder??

 

When I attempt bridging each port, it only recognises br0 and cannot see br1. br0 appears as custom br0 under docker network configuration. I've read up as much as I can. I have deleted my network configuration, tried setting my br0 as a static ip address, and br1 with no ip address so that it can be added in each docker instead, as suggested in a separate post. I've also tried more typical scenarios of assigning br1 with an ip address, etc. What isn't clear is that the second ip address cannot be assigned dns and gateway settings separately to br0. Not through the GUI anyway. The most I can get it to do is come up with an automated route of 'link'. I'm assuming to use br0 as a gateway to the router's gateway. Have I got this part wrong?

 

Currently, I've just deleted my network configuration and am starting again.

 

Any help would really be appreciated.

 

Edited December 22, 2019 by Geck0