I searched on this topic, as I have in the past, and only see hits from ‘the early days’ of Network Manager (2009-2011).
Having just successfully upgraded to 13.2/KDE, decided to check again.
No doubt Network Manager has matured over the years and runs very well on my laptop.
I have always used ifup for network interface management on my stationary machines, because way back when it was ‘preferred’.
Aside from legacy, is that still considered to be true?
Motivation:
I like to use NFS shares on my machines whenever possible, as some applications (in the past at least) did not properly interface to Samba shares when interfacing with the KDE file selector, and NFS always seems easier/faster (fewer navigation steps).
Network Manager is very tolerant of NFS shares that are not on line (silently does not mount them, quickly), whereas ifup will wait 5 minutes per share on boot-up, not very friendly.
I’ll probably just try it, but thought first I would ask about any updated philosophy on the topic.
Strictly speaking, “ifup” is gone. It is replaced by “wicked”, and some folk are unhappy about that.
I’m using “wicked” on my desktop, and “NetworkManager” on my laptop.
For most uses, it won’t make a lot of difference. Using “NetworkManager” gives individual users the ability to connect/disconnect. With “wicked”, you need root access for that. So I guess a server should use “wicked”.
Hmm, I just tried to make the switch on my 13.2/KDE desktop from wicked service to Network Manager.
It did switch to a Wired Connection 1", which appears to be DHCP, NM calls it Automatic.
But, the new Network Manager Connection editor is a lot different than 13.1, unclear how to create a static ipv4 setup.
That’s what I thought/expected as well.
Alas, the only fields not available when selecting Manual are the desired IP address and the Subnet Mask (I am comparing the setup tabs between 13.2 and 13.1.)
Based on Google search, I see folks over at Arch asking the same questions
I sometimes apply more than one addy to an interface when configuring devices (connected to a common switch for example). Then I can reach them both concurrently while configuring. There’s probably other valid use cases, but for most probably not ever necessary.
Thanks, Deano.
The only time I might need something like that would be when setting up new routers, which usually come set with 192.168.1.1 by default.
I change them to 192.168.10.1, mostly to differentiate “stock” from my setup.
I did not realize an interface could have two (or more) unique addresses.
So now I understand the NM setup scheme.