Ive been using linux since the 90’s but dont work on it enough to “know it” well. I install, configure it to a point and away to serving it goes. Anyhow Ive been using Fedora server for 4 or 5 years and fedora core before that. My expertise with it is extremely limited.
Equipment: Dell Poweredge R640 server. 256g 10-2.5 drives in a RAID0 quad gigabit ports
Anyhow… the problem.
I was running server 40 and then to 41. I system upgraded to server 42 and all was configured and good. Install dnsmasq and away it went. Never really configured nat/ipforwarding. Just set dnsmasq and it seemed to auto-configure. I always just used the dropdown in network card config installer when installing the OS and set my eno2 card to “shared to other computers” and completed the install. It got an update or something and I noticed my network cards started dropping. Wiped the whole thing to do a full clean install of server 42 instead of the upgrade from 41. Now when I use the dropdown to set it to “shared to other computers” it will take the configuration but wont connect. It shows connecting and after 30 seconds it stop and tries again. 30 seconds later it tries again and just shuts off. If I switch to “manual” on the dropdown it connects right away and shows eno2 network config data and connected. Ive tried everything and it wont let me use the “shared to other computers” option. I dont know whats changed? I tried rocky linux 9.5 and 9.6 and it was the exact same thing. If I just set it and then continue the install it was hit or miss whether it would share to other computers. When it did and all was connected if you reboot it just wiped all of the data from eno2 like it wasnt there. Weird. Ive been doing it this way for years but it just wont do it now. Something has changed. I even rolled back the firmware in IDRAC to see if it got a corrupted firmware update… nope it isnt that
I got creative. I figured something has changed. I installed Fedora server 42 with eno1 as external static ip (as always) and eno2 as internal static on 192.168.1.2. Completed install and both cards are there and configured. Do an update and then install gnome 48 on it and configure the network card eno2 in gnome as “shared to other computers”. It takes it fine and network is up and running again like before… however now everything is running on the 10.0.0.0 ip range instead of 192.168.0.0 ip range.
Simply put I cant find how machines are getting the ips? Did something in networking change with the release of server 42? I cant find what file/folder is issuing the dhcp leases. Where can I find it to edit it? I have devices running static ip in the 192.168 range. Im not a fan of the 10.0.0.0 ip range. Its going to be a whole lot easier to change the lease range back to 192.168 range than try to reconfigure all of these devices.
I dont get it???
Like I said. Im a 25 year newby that knows enough to be dangerous :-)~
The settings are stored in the NM’s connection profile.
Try using the CLI as described above, it works for me just fine.
I suspect the problem is related to the GUI config manager.
I checked the NetworkManager folder. Theres nothing in that folder anywhere thats assigning ip’s. I did follow the procedure you stated. It did change the ip and shut down internet connections on all other computers. Rebooted and it reset to the 10.42.0.1 IP again.
Im not getting this. I dont think I fully understand how this whole NetworkManager thing works. I know I read that it has a dnsmasq plugin installed on it that cant be used with “systemctl” command.
I try a few different things Im finding on the web but theres always one step that doesnt match up.
Im just lost and shooting in the dark here
NM launches dnsmasq with flags matching the configured parameters:
Failing to make persistent changes to the config may indicate filesystem corruption, or lack of free space, or a broken SELinux policy that can be ruled out with permissive mode.
Ok
Tonight I fired up my older poweredge r710. I installed fedora server 42 on it. During the install I used the shared to other computers method I have used before on eno2. Upon reboot every device on my network was up and running with an ip in the 192.168.1.XXX range. Running great without any problems.
I wiped it and did a reinstall of Fedora Server 42 again. This time using just manual on eno2 and rebooted. eno 1 was connected but no devices could get out through it. I installed and opened up nmtui. Configured eno2 from manual to shared and rebooted. All devices on the network are again up and running and able to get to the internet with ips in the 192.168.1.xxx range.
I dont get it?
Would a dell firmware update make a difference on whether it would allow me to share the connection on the r640?
There’s not enough data to draw any conclusions.
Try reproducing the issue on a fully upgraded F42 machine.
Collect the relevant configs, logs, error messages, etc.
