[Samba] Is Samba unable to resolve secodary group membership?

Fri Oct 9 10:00:50 UTC 2020

Hi all,

i read the logfiles again and again and stumbled over some lines:

[2020/10/07 11:25:45.191784,  5] 
../../libcli/security/security_token.c:63(security_token_debug)
   Security token SIDs (38):
     SID[  0]: S-1-5-21-3542048200-3079820972-537594794-55128
     SID[  1]: S-1-5-21-3542048200-3079820972-537594794-513
     SID[  2]: S-1-5-21-3542048200-3079820972-537594794-211797
     SID[  3]: S-1-5-21-3542048200-3079820972-537594794-92780
     SID[  4]: S-1-5-21-3542048200-3079820972-537594794-214631
     SID[  5]: S-1-5-21-3542048200-3079820972-537594794-5516
     SID[  6]: S-1-5-21-3542048200-3079820972-537594794-123946
     SID[  7]: S-1-5-21-3542048200-3079820972-537594794-73686
     SID[  8]: S-1-5-21-3542048200-3079820972-537594794-101266
     SID[  9]: S-1-5-21-3542048200-3079820972-537594794-84994
     SID[ 10]: S-1-5-21-3542048200-3079820972-537594794-58615
     SID[ 11]: S-1-5-21-3542048200-3079820972-537594794-62264
     SID[ 12]: S-1-5-21-3542048200-3079820972-537594794-73690
     SID[ 13]: S-1-5-21-3542048200-3079820972-537594794-211816
     SID[ 14]: S-1-5-21-3542048200-3079820972-537594794-63615
     SID[ 15]: S-1-5-21-3542048200-3079820972-537594794-75305
     SID[ 16]: S-1-5-21-3542048200-3079820972-537594794-211815
     SID[ 17]: S-1-5-21-3542048200-3079820972-537594794-211804
     SID[ 18]: S-1-5-21-3542048200-3079820972-537594794-211820
     SID[ 19]: S-1-5-21-3542048200-3079820972-537594794-211818
     SID[ 20]: S-1-5-21-3542048200-3079820972-537594794-22920
     SID[ 21]: S-1-5-21-3542048200-3079820972-537594794-92746
     SID[ 22]: S-1-5-21-3542048200-3079820972-537594794-211805
     SID[ 23]: S-1-5-21-3542048200-3079820972-537594794-92828
     SID[ 24]: S-1-5-21-3542048200-3079820972-537594794-73088
     SID[ 25]: S-1-5-21-3542048200-3079820972-537594794-211799
     SID[ 26]: S-1-5-21-3542048200-3079820972-537594794-169945
     SID[ 27]: S-1-5-21-3542048200-3079820972-537594794-211819
     SID[ 28]: S-1-5-21-3542048200-3079820972-537594794-128864
     SID[ 29]: S-1-5-21-3542048200-3079820972-537594794-101268
     SID[ 30]: S-1-5-21-3542048200-3079820972-537594794-128934
     SID[ 31]: S-1-1-0
     SID[ 32]: S-1-5-2
     SID[ 33]: S-1-5-11
     SID[ 34]: S-1-5-32-545
     SID[ 35]: S-1-22-1-20597
     SID[ 36]: S-1-22-2-10000
     SID[ 37]: S-1-22-2-10000001
    Privileges (0x               0):
    Rights (0x               0):
[2020/10/07 11:25:45.191945,  5] 
../../source3/auth/token_util.c:866(debug_unix_user_token)
   UNIX token of user 20597
   Primary group is 10000 and contains 1 supplementary groups
   Group[  0]: 10000001

If i read the lines correct, the S-1-5-21 sids are the ones which come 
from the ads. The SIDs starting with S-1-22 are the ones which are build 
by the unix user and unix groups the user is in. So it seems to me, that 
samba doesn't read the unix group memberships while building this 
security context. Is this behavior correct?

Unix user 20597 has a primary group id 10000 and 27 supplementary 
groups. None of these groups has an id of 10000001. Beside of this, 
shouldn't these groups also appear in the security token / unix user token?

Regards,
Michael

Am 08.10.20 um 11:31 schrieb Rowland penny via samba:
> On 08/10/2020 10:23, Michael Schwarz via samba wrote:
>>
>>
>> Am 08.10.20 um 10:41 schrieb Rowland penny via samba:
>>> On 08/10/2020 08:51, Michael Schwarz via samba wrote:
>>>
>>>> The setup at our university is not quite trivial. I can understand 
>>>> that. I'll try to explain it again in a different way:
>>>
>>> Lets see if I understand this, you have one kerberos domain for the 
>>> Linux machines and another kerberos domain for the Windows machines, 
>>> you have virtually the same users and groups in both. Why two 
>>> domains, why not just use the AD for both ? This would make your 
>>> setup trivial. I feel this is probably all down to department politics.
>>>
>>
>> Yes this is correct. I'm not sure why there are two domains. I'm not 
>> working at the central computer center, but i'm sure, they have their 
>> reasons why they are doing it this way. We are only using this 
>> infrastructure. The LDAP is storing much more information than only 
>> simple posixAccounts. It might be, that an AD is not so flexible if 
>> you want to store more than the standard attributes. But i don't now 
>> in detail as i am not so familiar with windows ad services.
>
> There are no posixAccounts in AD, there are just Accounts (but all the 
> RFC2307 attributes are available, so any account can be a Unix 
> account) and you will be surprised just how extendable the AD schema 
> is. No, I think it is just down to politics, Windows versus Linux 
> politics :-)
>
> Rowland
>
>
>

-- 
// Michael Schwarz - Universität Paderborn - PC2
// O2.152 - Warburger Str. 100 - 33098 Paderborn
// Telefon: +49 5251 601728 - Fax: +49 5251 601714
// E-Mail: schwarz at uni-paderborn.de