Le mercredi 19 juillet 2006 à 09:26 -0500, Aleksandar Milivojevic a
écrit :
> Quoting Adam Tauno Williams <[EMAIL PROTECTED]>:
> 
> >> Are there any recomendations should I use posixGroup or
> >> groupOfUniqueNames for new installations?
> >
> > Neither! Use "groupOfNames"; "groupOfUniqueNames" is not what you think
> > it is.
> 
> Hmmm...  Interesting, searching via google mostly returned references  
> suggesting most of the folks out there (and therefore tools they are  
> using) utilize groupOfUniqueNames.  However, I might be wrong.
> 
> Anyhow, if using either groupOfNames or groupOfUniqueNames, how about  
> gidNumber attribute from posixGroup?  I guess nss_ldap is not going to  
> work without it.  What would be the best way to add that attribute?   
> Other than defining my custom object classes or using extensibleObject  
> (obviously you do not recommend those two approaches)?
> 

Not quite. You should have a look at rfc2307bis (obsolete now but a new
version in preparation). Here is an old url

 http://www.padl.com/~lukeh/rfc2307bis.txt

I have a copy of draft-howard-rfc2307bis-01.txt from 20 February 2005 if
somebody is in the need. There posixGroup is defined as an auxiliary
class which can live with groupOfNames (groupOfUniqueNames) as in:


dn: cn=aaa,ou=groups,...
objectClass: top
objectClass: groupOfNames
objectClass: posixGroup
cn: aaa
gidNumber: xxx
member: uid=yyy, ... 
member: ... 


and nss_ldap will show uid yyy as member of posix group aaa with gid
xxx . Its just that rfc2307bis specifies that we should use
groupOfUniqueNames as structural class. May be Luke Howard should be
asked why it is so.

-- 
Marcel de Riedmatten





Attachment: signature.asc
Description: Ceci est une partie de message numériquement signée

---
You are currently subscribed to [email protected] as: [EMAIL PROTECTED]
To unsubscribe send email to [EMAIL PROTECTED] with the word UNSUBSCRIBE as the 
SUBJECT of the message.

Reply via email to