LUM, integrated login, roaming profile

Question

Re-reading many message threads in this and other forums, I'm
confused. For SLED, LUM creates users accounts on-the-fly after eDir
login (like DLU on ZEN on Windows), the Novell client reads login
scripts and maps drives (like on Windows) but integration with LUM is
less than transparent, and there may or may not be plans to support
roaming profies, without which settings and user document folders are
stuck on each machine the account was created on, making it very
difficult to support without mass confusion.
I run a college network. If we are to have a chance to make people
interested in an alternative to Windows, all those functions must be
transparent to users (as it is with the Windows client and ZEN) --
they authenticate to the directory, their local account is created,
they get their network resources, and their roaming profile settings
and locations are as expected. They logout, their info is safe on the
network, their account is deleted. They go to the next class in a
different lab and repeat. This happens thousands of times per day in
30 different labs with students of wildly different capabilities and
expectations. But it just works.
This has been the Novell experience since the beginning (16 years for
us). It is still valid. Anything less than this makes SLED an
unworkable alternative. I'm looking for a way to make it workable. By
summer we'll have 1200 XP/SLED dual boot machines in labs and offices.
To make SLED a viable alternative we need those other features.
Otherwise, it's just a toy for the curious. Unfortunate.
So, here are my end of year questions:
Which of these fundamental functions are working with which others?
Which ones are in the SP beta plans?
Which ones are not being planned?
What is the process for helping the planners understand the needs of
the users out here?
Thanks.
Lyle

Answer 1

Mark,
Thanks for the comments. I actually don't care if it doesn't implement
the same way as Windows. I agree, that is not a good model. Copying
the profile is dumb. If the plan is to simply mount the home, that's
by far the best way.
I suspect that what we are working with is a terminology problem. If,
we define "roaming profile" to mean that the user sees what they
expect to see regardless of what machine they login to, that is all
that matters. The need to copy the profile on Windows causes all kinds
of problems and is a poor solution.
So, let me re-state what I feel our needs are: 1) the user, not having
a local account, authenticates to the linux workstation using eDir, 2)
gets all the network resources per network scripts and policies, 3)
their home "profile" looks like it should, with all settings in tact,
4) they logout and no residue is left on the workstation.
However that is accomplished, I don't care and I'm not wedded to any
Windows terminology or processes. Hope this helps.
Lyle
On Wed, 27 Dec 2006 22:37:41 GMT, Mark Robinson
<[email protected]> wrote:
>On Tue, 26 Dec 2006 19:55:27 +0000, luzuml wrote:
>
>> Re-reading many message threads in this and other forums, I'm confused.
>> For SLED, LUM creates users accounts on-the-fly after eDir login (like DLU
>> on ZEN on Windows), the Novell client reads login scripts and maps drives
>> (like on Windows) but integration with LUM is less than transparent, and
>> there may or may not be plans to support roaming profies, without which
>> settings and user document folders are stuck on each machine the account
>> was created on, making it very difficult to support without mass
>> confusion.
>>
>> I run a college network. If we are to have a chance to make people
>> interested in an alternative to Windows, all those functions must be
>> transparent to users (as it is with the Windows client and ZEN) -- they
>> authenticate to the directory, their local account is created, they get
>> their network resources, and their roaming profile settings and locations
>> are as expected. They logout, their info is safe on the network, their
>> account is deleted. They go to the next class in a different lab and
>> repeat. This happens thousands of times per day in 30 different labs with
>> students of wildly different capabilities and expectations. But it just
>> works.
>>
>> This has been the Novell experience since the beginning (16 years for us).
>> It is still valid. Anything less than this makes SLED an unworkable
>> alternative. I'm looking for a way to make it workable. By summer we'll
>> have 1200 XP/SLED dual boot machines in labs and offices. To make SLED a
>> viable alternative we need those other features. Otherwise, it's just a
>> toy for the curious. Unfortunate.
>>
>> So, here are my end of year questions:
>>
>> Which of these fundamental functions are working with which others?
>>
>> Which ones are in the SP beta plans?
>>
>> Which ones are not being planned?
>>
>> What is the process for helping the planners understand the needs of the
>> users out here?
>
>Ok, there is a beta for NCL that will start in Jan. Please sign up at
>beta.novell.com if you are interested.
>
>At the moment, I have no idea what will and won't be in there! So why the
>reply?
>
>Well, things work kinda different under Linux. The idea of copying a
>roaming profile is very alien. In fact a roaming profile is just such a
>bad idea(tm) that it should never have been let loose in the first place
>IMHO :-)
>With Unix/Linux /home is a filesystem that can easily be exported from a
>server, and mounted on all workstations. That way the users' home
>directories are always available, but never need to be copied anywhere.
>Now, this doesn't fit well with OES Linux right now, but I am hoping to
>work with Novell on the OES 2 beta to try to get some of this stuff sorted!
>
>If you are serious about implementing Linux you need to get to know what
>some of the major differences are, and not just try to do the same thing
>again :-)
>
>I hope this doesn't sound like I'm ranting at you, it's not meant to, I
>just get a little narked when people see really **** stuff (like roaming
>profiles) and want it implemented when there's already a much better
>solution...