I guess its an old statement, but if you look in a bigger picture. Cloud providers are using endless leasetime DHCP. Its just simpler to handle the IP outside of the OS.
Your statement is more true like in the 1990s, early 2000s where CPU and RAM was limited, but today its not the case anymore.
Simple example. My home network contains about a hundred devices. It’s a large area and several buildings connected by cable and radio bridges. Of course I have a lot of wifi access points, some of which are routers set up as access points. One of these access points went crazy (maybe it was updated incorrectly or some other failure). It so happened that during a scheduled reboot of the central router, there was no electricity in part of the buildings at night. When it appeared, this crazy access point turned on its own dhcp server, distributed its addresses in its segment and decided that it was the main gateway. As a result, a whole segment of the network was completely lost to “smart devices” even though everything looked like it was working. Thus, all dhcp-dependent devices were affected by this failure. At that moment only those devices in which ip, gateway and dns were manually set, worked correctly. Of course, this is not a normal situation, and this “too smart” router went to the local flea market, and its place was taken by one that will not be able to do anything on its own.
I’m sure that the use of dhcp is applicable to such devices as smartphones, tablets, TVs and other devices for which ip address is not important in most cases. Well, or those in which it is impossible to set the address manually (the same sensors and tuya outlets). In such cases you need to assign ip addresses in the router with binding to mac. But such things as servers should be configured manually. And the range available for dhcp should be limited to a small and reasonably sufficient segment.
I was looking everywhere for exactly this. I was caught in a terrible loop where HA wouldn’t get an IP address so it couldn’t reach the time server, so it didn’t boot properly, etc.
I managed to change my IP address via CLI using the following commands:
network info to find the correct network adapter and see your current IP address set for HAOS
network update eno1 --ipv4-method auto --ipv6-method disabled (“eno1” needs to be replaced by your network adapter)
network update eno1 --ipv4-gateway 192.168.x.x (“eno1” needs to be replaced by your network adapter, “x” needs to be replaced by your IP gateway, probably router IP)
After doing this, my Fritz!Box auto-assigned showed an IP address for Home Assistant again.
This is real magic. I was trying to solve the issue for last three hours with no luck. Tried all options nothing has worked. Finally with just two commands, able to resolve the issue. Thanks a ton for this instruction.
Quite. Help is only help if it helps and I often find once I can’t resolve something from the GUI (where ordinary people can explore and learn), and have to drop to the CLI where I’m often in a whole world of hurt and frustration as I’m not a guru don’t speak their often highly abbreviated and full of assumption ‘help’ language. ;-(
‘Just type xyz’ when there might be 10 steps you have to have in place before you can do that.
Like whilst suggesting a new HA owner (Generic X64-86 HAOS) change his IP from DHCP to static, we ended up with a /32 netmask and so have locked ourselves out.
I have been trying to see what you might need to do to get that sorted from the CLI (via my own Odroid N2+ HAOS install) but nothing seems to work (the tips here seem to have got the closest) and so I think we are just going to re-flash the m.2 drive with a fresh HA and try changing the IP address again, before even restoring the one backup he took locally.
Mucking about with different SSH / Terminal tools, SSH keys, CLI logins and incantations that don’t seem to follow the guides …
Why couldn’t there be a HA splash screen terminal / menu driven way of setting such things, like you see of the RPi (Raspi-config or whatever it’s called)?