On Mon, 2006-06-19 at 18:39 -0400, G. Matthew Rice wrote:
> That _is_ the general idea that came out of the market research.  However,
> the interest and focus was highly concentrated on Samba and LDAP.

And as we've discussed before, I have no problem with those two topics
getting a majority share.

God knows that Samba configuration is still far more complex than even
an advanced, proper Kerberos/LDAP NFS setup.  And that includes all the
added non-file services to guarantee authentication, object mapping,
etc... to the enterprise-wide object stores (LDAP or otherwise).  Why?
Because we're emulating, bridging and doing all sorts of things.

But I do very much have a problem if we do a Samba-only exam, and don't
integrate Samba components into the greater auth/dir/name aspects, as
well as show where NFS also uses them.

Windows client authentication and object mappings (be they via Winbind,
LDAP, Kerberos and/or other mechanisms) and basic DNS and Windows name
resolution have _little_ to do with the Samba file services.  At the
same time, the concepts of the former have _everything_ to do with
general UNIX/Linux (among other system) network-wide services and
support.  Things that will need to be revisited for Apache, Database,
etc... as well, and not just NFS.

> Bryan, I do like where you're going with this.  Could you write something
> more concrete on these scopes somewhere up on the wiki?

Yes, I was planning to do so "all-inclusive" on the ELResource Wiki
first, then let people pick'n choose what doesn't and doesn't apply.
E.g., the ELResource Wiki will have an eDirectory section for its RSA
authentication, LDAP store, dirXML export, etc..., but we're not likely
to visit that at all because it's Novell-centric.

Got busy with a client all weekend, but sans one day this week, I should
have some good hours to devote to the outline and areas I believe need
to be covered.  Again, from there, I'll let others decide -- but I want
to make sure people are aware of where I'm driving it.

> I'm not certain if there will be a core exam for level 3.  If there is,
> though, (and it has been discussed) it would be the authentication exam.

In reality, we can call it whatever.  My point has been, and continues
to be, whatever is used by virtually everything should be LPI 301.  In
my experience, that's basic naming, that's centralized (possibly
distributed) authentication, and that's also how you publish resources
(be it broadcast, legacy resource listings, newer directory schema,
whatever).  Take the most common details and practices and focus on
those.

E.g., Winbind is heavily used not just for Samba, but also for Apache,
Proxy and other authentication -- even NFS and other mechanisms as well.
At the same time, Winbind doesn't have to be used at all, if your
UNIX/Linux enterprise is not talking to ADS, or is using another sync
option to ADS (like PassSync).

I'm kinda shocked my viewpoint is not understood.  But I'm commonly
known for not getting my point across in e-mail (far better in person,
even better in presentation with charts).  Or maybe I'm not seeing it
right.  I really don't know at this point, just kinda shocked.

> You nailed it, AFAIC.
> I think that everyone helping here has to step out of the 'this is the way I
> set up XYZ' and think about the bigger picture and all of the possibilities
> with these technologies.

I know everyone is concerned that we'll get "too generic."  But in
reality, there's not too much that is commonly used.  I hit on most of
them in one of my earlier posts.  Those capabilities are used for so
many things, not just Samba.

In fact, I'm kinda scared of putting just a few things in the Samba
exam, because they are really only used for standalone departmental
servers or, worse yet, when your enterprise is run by ADS.  We really
need to be documenting how you built an enterprise with UNIX/Linux as
your authority on your network (and not just standalone).

> The reason that we're picking things like Samba and LDAP is because they are
> big, complex and important technologies.  We want people that pass these
> exams to be knowledgable about these complexities.

Again, if we want to call the sames LDAP and Samba, so be it.

But I'm definitely not in favor of making the LDAP exam only LDAP, and
the Samba exam only Samba.  You have to not only address so much UNIX
filesystem and authorization detail in the Samba exam (which instantly
and simultaneously touches on NFS), but non-file services provided by
Samba like winbindd and nmbd also deal with clear auth/dir/name aspects
that are not independent of other network and system services.

> Those guys will be happy with LPIC-1 :)

My point exactly.  ;->

> You wouldn't be completely wrong in your assumption.  Luckily, the marketing
> folks aren't on this list.  As long as we cover the 'important' things to
> them (which is the enterprise level use of these technologies), we're good to
> formulate this in any way we think works best.

If I'm wrong, I'm not seeing it.  Or maybe I'm being too "absolutist."
I'm really trying to see it.  I just don't at this point.

> True.  This list is more to see what is in people's interest.

"Availability and Redundancy" is where I had many things go ...
- Storage, organization (disk labels, filesystems, LVM2, DM2, MD,
etc...)
- Redundant network filesystems  (shared/multi-target, GFS, etc...)
- Network-based availability
- Etc...

> BTW, weren't you working on an all encompassing proposal on LPIC-3
> developement?  Whatever became of it?

It's now going into the Wiki.  Just haven't had time to fill it in.  But
I will get some things in this week.  I had to track down my lawyer
today for other events (long story -- and not my current client ;-).

The proposal covered a lot of management aspects that are irrelevant.
And I had only gotten into the first 3 sets of objectives --
auth/dir/name, file/print and security.

> hmmm, LPIC-4...  Of course, how do you know when you've passed them all.
> We may add another one.

I've really avoided getting ahead of myself.  I've really just been
fixed on making sure we have a foundation for enterprise auth/dir/name
since everything else is based on it.

>From there, I had my "8 domain" EL model (which I literally wrote right
after the meeting of the 2nd day at LPI) -- not including development,
desktop or other things -- just for "network services."  That's largely
just so I have strict separation of services that makes the domains
manageable for myself.

> Agreed.
> Keep in mind guys, that when it comes time to do the JTA, all the tasks will
> be lumped together.  It'll be up to us, in the early fall, to combobulate
> them back into two real sets of objectives/exams.  We still have lots of
> wiggle room provided that we come up with an encompassing list of tasks for
> people to vote on.

And that's what I do agree is a good plan.

I still think it will come out into separating them out in how I'm
laying out the "8 domains" EL model, with many subdomains of concepts,
practices and tasks under each.

I'll fill those in shortly on-line.


-- 
Bryan J. Smith           Professional, technical annoyance
mailto:[EMAIL PROTECTED]     http://thebs413.blogspot.com
----------------------------------------------------------
The existence of Linux has far more to do with the breakup
of AT&T's monopoly than anything Microsoft has ever done.


_______________________________________________
lpi-examdev mailing list
[email protected]
http://list.lpi.org/mailman/listinfo/lpi-examdev

Reply via email to