samba - Mar 2019 - Is RODC password replication different from the windows version by design or is it a bug?

On 03/28/2019 05:32 PM, Rowland Penny via samba wrote:

[...]
>> Should the samba RDOC act like the windows version or is it different
>> by design?
>>
> 
> Yes it should and there is a bug report for something similar already,
> see here: https://bugzilla.samba.org/show_bug.cgi?id=13377
> 
> I know that is for members of the denied group, but the substance is
> the same, users are not getting authenticated on a RODC from a RWDC.
> 
> Can you please add to that bug report ?
> 
> Rowland
> 
> 
Thanks Rowland, that's exactly the topic. Garming Sam has commented it 
yesterday, the issue is that kerberos forwarding isn't implemented for 
now. That is exactly what wee seeing, authentication works __after__ 
(from the second attempt on) the initial password sync is done, the 
first attempt isn't proxied.

Adam

On Fri, 2019-03-29 at 10:16 +0100, Adam Minski via samba
wrote:
> 
> On 03/28/2019 05:32 PM, Rowland Penny via samba wrote:
> 
> [...]
> 
> > > Should the samba RDOC act like the windows version or is it
different
> > > by design?
> > > 
> > 
> > Yes it should and there is a bug report for something similar already,
> > see here: https://bugzilla.samba.org/show_bug.cgi?id=13377
> > 
> > I know that is for members of the denied group, but the substance is
> > the same, users are not getting authenticated on a RODC from a RWDC.
> > 
> > Can you please add to that bug report ?
> > 
> > Rowland
> > 
> > 
> 
> Thanks Rowland, that's exactly the topic. Garming Sam has commented it 
> yesterday, the issue is that kerberos forwarding isn't implemented for 
> now. That is exactly what wee seeing, authentication works __after__ 
> (from the second attempt on) the initial password sync is done, the 
> first attempt isn't proxied.
It should work, as long as you are using the internal Heimdal KDC, and
I thought we even had tests for that.  The KDC propagates up a special
error code to the processing layer to say 'please proxy this packet to
a full DC' to trigger that

There are other things we don't fully implement (like forwarding bad
passwords, we do that by sending a bad NTLM password, not a Kerberos
one), but this much should work...

Andrew Bartlett

-- 
Andrew Bartlett                       http://samba.org/~abartlet/
Authentication Developer, Samba Team  http://samba.org
Samba Developer, Catalyst IT          http://catalyst.net.nz/services/samba

On 03/29/2019 10:37 AM, Andrew Bartlett wrote:
> On Fri, 2019-03-29 at 10:16 +0100, Adam Minski via samba wrote:
>>
>> On 03/28/2019 05:32 PM, Rowland Penny via samba wrote:
>>
>> [...]
>>
>>>> Should the samba RDOC act like the windows version or is it
different
>>>> by design?
>>>>
>>>
>>> Yes it should and there is a bug report for something similar
already,
>>> see here: https://bugzilla.samba.org/show_bug.cgi?id=13377
>>>
>>> I know that is for members of the denied group, but the substance
is
>>> the same, users are not getting authenticated on a RODC from a
RWDC.
>>>
>>> Can you please add to that bug report ?
>>>
>>> Rowland
>>>
>>>
>>
>> Thanks Rowland, that's exactly the topic. Garming Sam has commented
it
>> yesterday, the issue is that kerberos forwarding isn't implemented
for
>> now. That is exactly what wee seeing, authentication works __after__
>> (from the second attempt on) the initial password sync is done, the
>> first attempt isn't proxied.
> 
> It should work, as long as you are using the internal Heimdal KDC, and
> I thought we even had tests for that.  The KDC propagates up a special
> error code to the processing layer to say 'please proxy this packet to
> a full DC' to trigger that
We use the internal Heimdal KDC, and it doesn't work, at least for 
version 4.9.4. Is there any stuff I can test? Or can you give me an 
entry point to the code? Thanks.

Adam
> 
> There are other things we don't fully implement (like forwarding bad
> passwords, we do that by sending a bad NTLM password, not a Kerberos
> one), but this much should work...
> 
> Andrew Bartlett
>