Guillaume Rousse wrote on 10/07/2008 04:08 AM:
> Hello list.
> 
> Thanks to archives of this ML, as well as source code from
> Net::LDAP::Class::User::AD package, I'm able to create functional AD
> user entries from Net::LDAP. 

Any particular reason you didn't just use Net::LDAP::Class::User::AD
directly?


> Third, is there a recommended practice for organising user and group
> entries in AD ? In OpenLDAP world, the standard practice is to have a
> 'user' and a 'group' branch, whereas AD setup I saw sofar had a more
> subdivised organisation (one branch per group, for instance).
> 

I don't know of recommended practices; I can, however, speak to how
Net::LDAP::Class::User::AD does it and how we do it at $work, which are
different (despite the fact that I wrote Net::LDAP::Class).

NLCUA assumes you are using the primaryGroupID attribute in AD, since
that attribute is required for Unix compatability even though AD doesn't
itself use it. In other words, the Net::LDAP::Class basic API assumes
every user has a primary group and so the AD classes implement that API.
 AD has Security Groups which can be used like OpenLDAP groups, and that
is what Net::LDAP::Class::Group::AD assumes.

However, at $work we use OrgUnits with AD in a way analogous to how
OpenLDAP uses primary groups. We use Security Groups in AD to control
which resources a a User has access to (e.g., we have just a few
Security Groups, while we have 100s of OrgUnits). I actually have a
Net::LDAP::Class-derived class for OrgUnit and each User class uses a
OrgUnit class as its primary group.

I think the $work approach succeeds because of our particular security
requirements. I hope that Net::LDAP::Class::*::AD can also work for
folks implementing Unix-style organization for their AD server.

-- 
Peter Karman  .  [EMAIL PROTECTED]  .  http://peknet.com/

Reply via email to