On 5/1/20 10:00 am, Emmanuel Lécharny wrote:

On 04/01/2020 23:59, Brian Burch wrote:
On 5/1/20 2:45 am, Emmanuel Lécharny wrote:
Hi Brian,

it works for me, but I have added the MUST attributes homeDirectory and uid (from posixAccount):


objectclass ( 1.3.6.1.1.1.2.0 NAME 'posixAccount'
     DESC 'Abstraction of an account with POSIX attributes'
     SUP top
     AUXILIARY
     MUST ( cn $ uid $ uidNumber $ gidNumber $ homeDirectory )
     MAY ( userPassword $ loginShell $ gecos $ description ) )


Your LDIF does dot add those two mandatory attributes. the error message is not appropriate though...

Thanks for your quick and helpful reply, Emmanuel.

Yes, that was my own conclusion. However, the best of my google searches ALL turn up with the same ldif that worked successfully under the M23 schema.


Do you mean that this ldif worked with M23 ?

Yes!

Here is the ldapEntry copied with studio from my own M23 directory which is running in production:-

dn: m-oid=1.3.6.1.1.1.2.0,ou=objectClasses,cn=nis,ou=schema
objectclass: top
objectclass: metaTop
objectclass: metaObjectClass
m-oid: 1.3.6.1.1.1.2.0
m-description: Abstraction of an account with POSIX attributes
m-may: userPassword
m-may: loginShell
m-may: gecos
m-may: description
m-must: cn
m-must: uid
m-must: uidNumber
m-must: gidNumber
m-must: homeDirectory
m-name: posixAccount
m-obsolete: FALSE
m-supobjectclass: top
m-typeobjectclass: AUXILIARY

Obviously, all the MAYs and MUSTs are defined. I have no evidence or memory of me having hacked this schema myself in the past. I'm pretty sure I just enabled it and then created the atomic ldif to add the the extra objectClass in the same ldapmodify as adding the attribute values. I keep this ldif in my source repository.

Obviously, one of our devs updated cn=nis,ou=schema in either M24 or M25, so now the advice on the internet is both wrong and confusing..

Can you proivide the pointers on the internet pages you are mentionning?

I found quite a few which were related, but this was the most explicit:-

https://stackoverflow.com/questions/52604987/apacheds-gidnumber-not-declared-in-objectclasses

Note particularly the answer section, which states "Since the objectClass has mandatory attributes, this needs to be performed as a single operation."

The PosixAccount ObjectClass hasn't been changed since 2011...

Wow! I don't know what to say! I installed M25 from git quite recently and it must have come with the "broken" schema because I didn't change it.

How do you explain the fact that you needed to hack your own local schema?

Clearly we need to figure out what is going on here.

Do you agree the nis schema from 2011 is also the "broken" version?

Do you know how our project generates this schema? Is it automatically cloned from some external repository, or simply maintained manually?

Puzzled and not able to make any progress..

Brian

Can you shed any light on the reason for this change? I though the registered owner of the nis schema OID space made the decision about its content, while the apacheds project merely accepted any changes and (of course) is free to maintain the m-disabled switch as it wishes...

I don't have my new M25 system running at the moment, but am I right in thinking this current M25 schema makes the posixAccount objectClass almost pointless? It's only use would be as a search term.

I can't see why ANY ldap object at all could not have the uidNumber, etc, attributes assigned to it once the nis schema has been enabled? I haven't tried yet, but these ldap attributes don't seem to be defined in any objectClass that I could find...
uidNumber AT is used by the posixAccount, sambaUnixIdPool and sambaIdmapEntry ObjectClass.

WDYT? I don't want to change the schema back to M23 (like you did!) without a good reason, because simply circumventing the current problem doesn't feel right to me. There's the matter of the misleading google searches, too!


Again, there was no schema change made since a very long time, and certainly not between M23 and M25.


There must be something else that cause your issue (and again, I did check on my current trunk, and the ObjetClass change just works).



Regards,

Brian

On 04/01/2020 07:28, Brian Burch wrote:
I am currently trying to do a clean setup of a new M25 installation using and adapting the set of ldapmodify jobs I successfully used many years ago for my production M23 system (which still works fine).

I've hit quite a few issues with additional schema elements, so I decided to take it slow and easy to just get the posix stuff working first. It has been driving me pretty crazy and I really can't work out what is wrong.

Of course, I have modified:-

dn: cn=nis,ou=schema
changetype: modify
replace: m-disabled
m-disabled: FALSE

... and restarted the server. Studio confirms my change is active.


Here is the existing basic user definition (copied as ldif with studio):-

dn: uid=brian,ou=people,o=pingtoo.com
objectclass: inetOrgPerson
objectclass: organizationalPerson
objectclass: person
objectclass: top
cn: Brian E. Burch
sn: Burch
givenname: Brian
uid: brian
userPassword:: e1NTSEF9WWsvOXZhN3ZtNkxBemhYeURmOEczNjhPSjJndGkwazNJeVphelE9P
 Q==


So now I try to make a very simple change to turn my ldapentry into a posixUser:-

dn: uid=brian,ou=People,o=pingtoo.com
changetype: modify
replace: objectClass
objectClass: top
objectClass: person
objectClass: organizationalPerson
objectClass: inetOrgPerson
objectClass: posixAccount
-
add: uidNumber
uidNumber: 2000
-
add: gidNumber
gidNumber: 2000


But it fails as follows:

ldap_initialize( ldap://ldap.pingtoo.com:10389 )
replace objectClass:
    top
    person
    organizationalPerson
    inetOrgPerson
    posixAccount
add uidNumber:
    2000
add gidNumber:
    2000
modifying entry "uid=brian,ou=People,o=pingtoo.com"
ldap_modify: Object class violation (65)
    additional info: OBJECT_CLASS_VIOLATION: failed for MessageType : MODIFY_REQUEST
Message ID : 2
    Modify Request
        Object : 'uid=brian,ou=People,o=pingtoo.com'
            Modification[0]
                Operation :  replace
                Modification
objectClass: top
objectClass: person
objectClass: organizationalPerson
objectClass: inetOrgPerson
objectClass: posixAccount            Modification[1]
                Operation :  add
                Modification
uidNumber: 2000            Modification[2]
                Operation :  add
                Modification
gidNumber: 2000org.apache.directory.api.ldap.model.message.ModifyRequestImpl@4477d5e7: ERR_277 Attribute gidNumber not declared in objectClasses of entry uid=brian,ou=People,o=pingtoo.com

modify ended with rc  65


So I take a look at cn=nis,ou=schema..

* the attributes m-oid 1.3.6.1.1.1.1.0 uidNumber and 1.3.6.1.1.1.1.1 gidNumber are defined.

* the objectClass m-oid 1.3.6.1.1.1.2.0 posixAccount is defined as structural, metaTop, BUT it does NOT have MAY or MUST entries for uidNumber or gidNumber!

* I have searched all the other nis objectClasses, but can't find any reference to the attributes I want to associate with my user. I can't find the less important ones either, e.g. loginShell, homeDirectory or gcos.

It has been a long time since I last did this kind of basic setup, so I accept I have probably overlooked something simple and obvious.

HOWEVER... when I look at the M23 schema, objectClass=posixAccount has MUST for uidNumber, gidNumber and homeDirectory. It also has MAY for the other attributes I want to associate with this user. Why the difference?

I would be grateful if anyone wiser could let me know how to get past this show-stopper.

Thanks...

Brian

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to