I’m trying to get virtual guest OSes to network with the hypervisor host OS. That means fiddling with the network settings. A good thing to know in such a situation is how to start from scratch, without blowing the host OS away and reinstalling. Are there any files in, say, /etc/syscontrol/network/ that I should or should not remove? What about /dev/.syscontrol/network/?
Some related questions:
If I understand correctly, kernel modules for the netcards are automatically loaded unless they are blacklisted.
How does OS13.1 know which modules to install on the hard drive during the initial OS installation?
Will hwinfo detect hardware that has no corresponding modules installed?
First,
Although this is a networking question, it has everything to do only with setting up networking for virtualized guests so should have been posted in the Virtualization forum. The problem, related issues and solution will have little to do with ordinary networking.
So,
Resolving this will require some understanding of how networking works with virtualized guests. Although it might be possible to shoehorn a Guest into using the raw Host NIC, that’s not typically done nowadays.
The correct procedure is to create a virtual networking device (Linux Bridge Device) on the Host, and then configure Guests to use the device for type of network you wish to run (Linux Bridge Devices can be setup typically as bridging (same networked as the physical network), NAT, HostOnly). Switching from one network to another is as simple as re-configuring the Guest properties to point to the appropriate LBD.
How you create your LBD will generally depend on the virtualization technology you’re using. If you’re using VMwar or VBox, their GUI managers do this easily. If you’re running anything else, there is a good chance that libvirt will give you similar functionality (vm manager, vm install, virt install). You can create very basic devices using YAST. You can also create them using the command line.
Recommend you first search the Virtualization forum for Q&A related to your basic question and if you still have a question, can create a new post there.
And as always, when you create an initial post, try to be as descriptive as possible. Be sure to mention your OS (and version), the virtualization technology you’re using (and version), post any output or errors in
blocks (the hash button in this Forum's text editor) and anything else you believe relevant.
HTH,
TSU
I want to get back to what YaST started with. I can go to the GUI and delete the configured devices, but I’m not sure that removes all of the changes I’ve made. I want to be sure I’m getting s fresh start.
I was asking about networking in general. I have started a thread on the virtualization forum to discuss what is actually happening there. the OS I’m using is listed in the subject line. Since this was a very general question, it doesn’t seem relevant to include additional information.
If your intent was about “networking in general” you didn’t frame it that way.
Based on your original post
/etc/syscontrol/network/ that I should or should not remove? What about /dev/.syscontrol/network/?
Directly responding to your above,
I may not understand what you are trying to say. To configure networking between a Host and Guests is as I described, through the creation and use of Linux Bridge Devices, not some general non-networking configuration.
So, specifically no…
There aren’t any settings in /etc/sysctontrol/network/ or /dev/syscontrol/network I know of that’s relevant to setting up networking between a Guest and its Host.
Some related questions:
If I understand correctly, kernel modules for the netcards are automatically loaded unless they are blacklisted.
How does OS13.1 know which modules to install on the hard drive during the initial OS installation?
Will hwinfo detect hardware that has no corresponding modules installed?
Today’s Linux subsystem udev does the hardware discovery, recognition and device driver and firmware “installation” and configuration. In plain words, if some hardware is detected, it is analyzed and a hopefully appropriate device driver is installed and configured to use that hardware. Just because kernel modules are loaded does not necessarily mean they are used, in general the kernel provides the capability but the OS has to decide to use the capability.
AFAIK hwinfo has nothing to do with what udev does, it only reports information but does not actively do anything.