I am new to linux, I have 5 PC's running SUSE 9, and a Server running SUSE 8.

All devices appear to be working normally and I can ping all of them, what do I need to use to login [I am not sure if this is the correct term in Linux] to the server from each of the PC's?

I need a home folder for each user on the server, the users need to login from any of the 5 PC's

What is the best way to achieve the above?

Many thanks

TheHammer

Hi, these are a few steps to begin:

in the server:
- add all the users needed.
- Make sure sshd (and other services like ftpd, httpd, ...) damon is installed and running to let the users log in the machine.

in the workstations:
- Make sure ssh packet is installed.
- ssh to log in server machine.

If you want the server to act as file server, then you can use nfs to share directories (editing /etc/exports) and mount them in the clients via /etc/fstab. See nfs-howto for details.

Bye.

Hi Goala

Thank you for your reply.

In the Linux world - does every PCrunning Linux need every user to be added locally? For example John Smith would need an account on PC1, PC2, PC3 etc. Jane Alway would need an account on PC1, PC2, PC3 etc. and so on? Is there a way of loging in to the server directly from the Login screen presented on the PC? Is there a method of login to the server and PC at the same time?

As you can tell I am very new at linux networking.

Is NFS the best way of file sharing on a Linux server?

Many thanks for your help

Hi again!.

You don't need to create the users in all the computers, it would be very laborious. You can use LDAP instead.

See
LDAP Authentication for Linux or LDAP Howto for details. I have no experience with LDAP so I'm not very useful here. I hope this can help you.

And about NFS, well I only know this to share filesystems transparently between systems (well and samba, but this is for Win-Linux). Once you have mounted a file system (or directory) of another machine in yours, it's transparent to you. The new directory is like anyone.

Bye and good luck!

PS: keep us informed of your progress.

I'm also a newb, and I can't seem to start sshd (on suse 9.2). It keeps saying sshd re-exec requires execution with absolute path. What's that? I also would like to know how to login to ssh . I can't find a good howto.

-Peter

"Absolute path" is (more or less) another way of saying "fully qualified pathname"; "./sshd" is a relative pathname (it's relative to your current working directory, or "."); "/etc/rc.d/init.d/sshd" is an absolute path, as it starts at "/".

ssh hostname

If you're logged in as one user on "host A" and want to log in to "host B" via ssh as a different user, it's:

ssh -l username hostname (that's a lower-case "L")

The "getting place" for howtos for most all things Linux is http://tldp.org/ ; I think the best you're going to do w respect to using ssh is "man ssh" though.

FWIW, particularly if you're new to this, I would strongly suggest starting with something a little more forgiving, like SAMBA, and leave LDAP for later. That's just my $.50 though.

Good luck.

- Larry

Thanks. Over the last few years I've learned a lot more about ssh; enough to know that SuSE has done something weird with it's ssh and it wants to be started via YaST2. I've always executed it from an absolute path, whether it gives me the error or not, and when I'm working with anything but a Fedora or SuSE machine it works fine.

-Peter

That's - interesting... As far as how ssh itself works, there should be no difference from implementation to implementation.

Are you referring to a "starting sshd from inetd" vs. "starting sshd from "/etc/rc[1-5]{1}\.d/S[0-9]{2}sshd" vs. "starting it from rc.local by way of a command such as '(csh -cf /usr/bin/sshd > /dev/null) >& /dev/null', etc.?

It sounds like what's confusing you - at least in part - is the bootstrap process; that is, how a Linux machine with a particular distro on it starts up. Specifically, I think what's confusing you is that there are differences in the way that different Linux distros initialize. Those differences range from "minor" to "complete and total". FWIW I would check the distro you're working with and fully understand whether it's doing a System V init, or something else...

How's the PCs & server network project coming, everything up and running?

- Larry

I'm aware of certain differences in the bootstrap process; for example on my distro I use /etc/init.d/sshd start (no rc.number, or at least I've never used it). I know that the bigger distros, like SuSE, Mandrake, and presumably RedHat like to mess with some of their packages to make them more secure or what have you. I haven't tried initializing ssh by hand in a while, I'll try it again when I get a chance.

So yours has its services set right up in /etc/init.d but doesn't do the entire Sys V init... interesting... I was messing around with "Rock Linux" the other day; the "where" they chose to put their scripts, and how they chose to start their services, from what I saw, can best be described as "all over" and "ad hoc" - in my opinion, for whatever that's worth.

We have our own "internal use only distribution", created out of necessity when RedHat changed some fundamental aspects of RPM and a few other things back at version 7; those changes screwed up the configuration (mix of RPM && software compiled from tarballs & installed), completely. So, we reinvented the wheel. However, about the only thing the world needs less than another programming language is another Linux distro. We use it when we set up turnkey systems, but that's about it. In some ways I regret holding it back, as what's happened to it, is that various people have added bits and pieces of other OSs we support, they've worked with, etc., so it's extremely usable. It's probably the only Linux distro with a JES2 facility and that provides xedit / kedit && ISPF...


Generally speaking, it's not security so much as it's compatibility; other *NIX systems fall into pretty much two categories: BSD and System V. The purpose for having a Linux distro do a System V init is for flexibility, but also, so that code can be easily ported from a *NIX system to and from Linux; for example, the startup and shutdown scripts for my mail server && DNS server are the same on Solaris and on Linux; I only had to write, and only have to maintain, one set.

The big argument I've heard from those who support other distros is that the System V init is rather cumbersome, not the easiest thing in the world for a first-time user to figure out, etc.; personally, I don't buy it. I learned it without much difficulty; it made sense to me, and, it got me thinking about how the whole OS fit and worked together.

Plus, if I go from one distro to another, and they're both using the same setup, I won't have to spend another hour trying to figure out where the hell they hid the network startup script and how in the world it works. Never thought I'd be the one saying this, but there's something to be said for compatibility, even if it's not the fastest or most effieicient thing around.

Yeah, open a console and fire it up in foreground mode. Or use "screen" to do something similar. It's always informative to see what, exactly, the OS daemons are doing.

You might want to consider the setup that this guy advocates:

http://cr.yp.to

Check out his "daemon tools" software. Basically, you create one "services" directory, and everything's controlled by a "supervisor" daemon that's started by inittab. I've found them to work very well.

- Larry