This is on openSUSE 13.2_64 both on a PC and a netbook. I want to connect PC and netbook just via ethernet cable (locally, no internet, like null-modem). Each machine therefore needs a static IP (if I’m not mistaken) but I found no obvious way to assign local static IPs in Networkmanager. What am I missing?
Which NM frontend are you using? (KDE’s plasma-nm, GNOME’s integrated one, nm-applet?)
Although they should all be similar anyway.
In KDE:
Open the connection editor, switch to the “IPv4” tab, select “Manual” as method, and a new field should appear. Click on “Add” and type in the IP-Address, Subnet mask, and Gateway.
Or just switch to Wicked and configure it in YaST.
There’s no need for NetworkManager if you only have a static Ethernet connection.
When you mean just a cable between the two, you are aware of the fact that you need a “cross” cable and not a normal one?
Ups, it is KDE’s plasma-nm.
Sorry to grouch, but this is not really an “obvious way” sigh. I’ll explain where its obstacles for a newcomer like me lie hidden: When I edit a connection, I expect to have to enter the destination’s IP. My own (on the machine I am working on) should in my naive way to look at things be set during boot or so. And for a simple cable I don’t need a gateway I presume. Can I leave that empty?
Is Wicked the software which takes over when I disable networkmanager in YAST? Is this a frontend to ifconfig? As to “no need” … well, in most cases one reads that networkmanager is the way to do networking since already several years time, anything else more or less deprecated.
Explanation 2: I wanted to copy files between the machines. I also have mobile internet on the PC and an UMTS-stick for the netbook, both of which were not suitable for this simple connection. Thank you for your input :).
Not necessarily.
IME, a normal one should work just as well.
Some (many? most?) ethernet interfaces detect whether it is a normal or “cross” cable and change their behaviour accordingly.
Yes, the cable works “as is”.
Why? If it is set to “Automatic”, you don’t have to enter anything. That’s the point of automatic.
And the default is to make it as easy as possible to connect to a standard LAN.
Your use-case is not the common one.
My own should in my naive way to look at things be set during boot or so.
They are, once you configured them how you need them.
And for a simple cable I don’t need a gateway I presume. Can I leave that empty?
Yes.
The gateway is only used/needed for “unknown” addresses, i.e. ones that are not in your “LAN”.
Is Wicked the software which takes over when I disable networkmanager in YAST?
Yes. It’s (open)SUSE’s new replacement for ifup.
Is this a frontend to ifconfig?
Well, not exactly.
“ifconfig” is deprecated since kernel 2.0, and has been dropped completely recently in Tumbleweed.
It is a frontend to the kernel’s network functions, just like NetworkManager, ifconfig, or ifup.
As to “no need” … well, in most cases one reads that networkmanager is the way to do networking since already several years time, anything else more or less deprecated.
Yeah, on Ubuntu maybe.
openSUSE always defaulted to ifup (unless it detects a Wireless interface during installation), and since 13.2 Wicked.
That said, if you often switch between different networks (different wireless connections, VPN, …), NetworkManager makes it easier.
But AIUI, that’s exactly the reason why SUSE developed Wicked, to “fix” those shortcomings in ifup, and make dynamic connections easier.
At the moment, only NetworkManager is supported by the desktops, you need to configure Wicked in YaST (or manually by editing the config files). AFAIK a desktop applet for configuring Wicked is planned, but it’s not available yet.
Automatic will work only with DHCP, won’t it? I can relate to a simple approach. My error was assuming that the most basical case (1 cable between 2 computers) is still the first step for everybody else and thus covered by the usual setup.
Ah, thanks.
Again, thanks.
I see.
So it is SUSE-specific then.
lol!
Thanks for the information.
I’ll try it out tonight. Thanks for your help.
Yes. It has to take the settings from somewhere…
And that’s the most common case.
A normal user expects to just plugin his ethernet cable into his system and the router/modem (or LAN) and have working Internet.
I can relate to a simple approach. My error was assuming that the most basical case (1 cable between 2 computers) is still the first step for everybody else and thus covered by the usual setup.
Using DHCP is standard since a long time.
From https://en.wikipedia.org/wiki/Dynamic_Host_Configuration_Protocol:
DHCP is based on BOOTP …
It was first defined in RFC 1531 in October 1993; but due to errors in the editorial process was almost immediately reissued as RFC 1541.
Even Windows 98 defaulted to DHCP already IIRC (don’t know about '95).
And if you don’t use DHCP, you have to configure the network anyway. But why should people using DHCP have to configure it? The whole point of DHCP is to avoid having to configure anything on the client side… 
So it is SUSE-specific then.
Yes. But it is somewhat “compatible” to the old ifup, e.g. you can still use ifup/ifdown (they get redirected to Wicked), and it reads the old config files (from /etc/sysconfig/network/) if available.
Ahmm, just to clear things a little more … when I assign static IPs to both machines they are “seen” by the respective other (along with any access to the internet)? So dolphin will have an entry in the places tab e.g. for netbook and I can see the files there (probably after entering user/password)? rsync and ssh would work at once?
Well, they are “seen” in so far, that you can reach the other via the IP address (provided that you set up both correctly).
The Internet access is not forwarded from one machine to the other automatically.
So dolphin will have an entry in the places tab e.g. for netbook and I can see the files there (probably after entering user/password)?
No.
If you have a Samba server running on the other host (or a Windows with shares), dolphin should show it under Network->Samba Shares.
Other ways to access it with dolphin would be to type something like the following into the location bar:
sftp://username@ip-address (needs sshd)
fish://username@ip-address (needs an sshd)
smb://username@ip-address (needs a Samba-Server, you could also use the hostname instead in this case)
There would also be other possibilities like FTP, for which you’d also have to run a corresponding server/daemon on the other host.
rsync and ssh would work at once?
Yes. If the other host runs an rsync or ssh daemon/server. (sshd should be enabled by default in openSUSE I think, but normally the Firewall blocks it)
Okay, got it. Thanks again.
On 09/10/2015 11:56 AM, wodenhof wrote:
> wolfi323;2727625 Wrote:
>> In KDE: Open the connection editor, switch to the “IPv4” tab, select
>> “Manual” as method, and a new field should appear. Click on “Add” and
>> type in the IP-Address, Subnet mask, and Gateway.
> Sorry to grouch, but this is not really an “obvious way” sigh. I’ll
> explain where its obstacles for a newcomer like me lie hidden: When I
> edit a connection, I expect to have to enter the destinations IP.
Well, if you depart from “automatic”, then you have to enter the address
somewhere. And as there are IPv4 and IPv6 addresses, you get two tabs,
one for each standard. Otherwise, you’d have a single dialog a foot long 
–
Cheers / Saludos,
Carlos E. R.
(from 13.1 x86_64 “Bottle” at Telcontar)
I wasn’t grouching because of the tabs but because the dialogue / menu is obviously there to set up a connection to another machine / net. From there it is not obvious that the very same menu / dialogue will assign a static IP to my local machine with no further reference to a network / second machine. IMHO.
Either I do not understand your stories att all, or you seem not to undertstand very much about TCP/IP networking. I applogize if the first is the case.
You do NOT specify another system on the system you are configuring. You configure an IP address and a netmask on your own system on a NIC. That determines which LAN that NIC is on. And it can then talk to every other system (NIC) on the same LAN (that is the physical LAN, which is the connection cable on your case, and the same network/netmask). Thus when you have defined the NIC in having 10.1.2.3/24 (the /24 is the same as netmask 255.255.255.0), the system can connect to any 10.1.2.0/24 system on that cable. Of course you should know that that other system is there. Thus when the system at the other end of the cable is 10.1.2.17/24, you will try e.g.
ping 10.1.2.17
to check the connection.
Thus you only define the IP address and netmask of your own NICs. Which is in the most cases done using DHCP, but in this case is to be done “manualy”.
BTW. I hope you did choose the IP addresses you use on that LAN (the cable with two ends) carefully. I hope this remark is completely redundant to you, but you can not use any IP address you happen to think of.
That is right, I am quite new to networking (actually these are my first steps). No need to apologize :).
Ahhh, bingo. Thankyou, you just nailed my central misconception. Funny, the way one (me at least) can read and still not understand. This is really helping a lot.
Yes. Indeed.
This I knew. I used (or rather tried to use) the addresses in the range reserved for private nets (192.168.x.x or 172.16.x.x).
Nice to be of help here. It is always difficult to see that one misses something, but not being able to get the message through because of the distance in time and space we have on the forums. I am glad I ssucceeded.
These are fine, but be sure you choose a different one from the other LANs you have. I assume that at least one of the systems we are talking about is connected through another NIC to another LAN (that may itself be connected to the internet). Such a LAN most probably also uses a private address range, often fixed in the router/modem from your ISP (they often use 192.168.x.x).
Next misconception squashed. I was thinking in terms of machines (PC, netbook, …) while I should have thought in NICs. This avoids the next trouble, because up to right now I planned to use on my netbook the same address for WLAN and ethernet.
Thanks again, Henk.
You are welcome.
I nice methaphore I sometimes use is postal addresses and postal packages (but as every methapore, it has it’s limts).
A LAN is the street. The NICs are the doors into those streets. Those streets are in fact dead alleys, but they can have doors into houses that have other doors into other streets. Routers not only have more doors, but also have a handpost to show through what doors to go to what addresses
When you have two doors, they (in this case) lead to different streets. Your internal handpost (routing table) sends packges through the front door for the the street there and through the back door for the other street. When packages have addresses different from both streets, they are send to the post office (the default route). You can have only one post office and thus it can only be in one of the streets. When packages arrive in the post office, they will be send further through another door of the postoffice to the next post office with your ISP. And then they will travel according to all sorts of routing tables to the address wanted.
One of your questions I interpreted as: when one of those systems (A) can go to the internet and I connect the other system (B) using the cable, is B then able to connect to the internet?
The above shows that that can be done by (this is the basics, no details):
- configuring the default route in B and let it point to A;
- A will not normaly forward packages, thus it must be configured to do so.
Another I interpreted as: when II have the TCP/IP connection between A and B, do then all sorts of applications automagicaly start talking to each other over the connection? No they don’t.
On the receiving system there must be a program listening on the appropriate TCP or UDP port, such a program is called a server. And when it is active waiting for a client, it is aslo often called a deamon.
On the other system, a client program must be started to connect to the server.
So you gave e.g. a SSH server when you run sshd. And the client then starts the client program ssh to connect to it.
BTW, when you have fixed those IP addresses, it might be nice for day to day working to connect a name to them. Thus the ping statement and all other calls to a server function could use the name:
ping thingy
To do this, put them in your /etc/hosts. Either using an editor or YaST.
Nice analogies :).
Good hint, too, to assign names. I will do that later on.
P.S.: Does this forum have the ability to add something like [solved] to the title, so other users can find actual solutions easier?