linked attributes, DRS and abusing the ldb modules API

[Stefan (metze) Metzmacher]

Hi Tridge,

>  > I don't understand that question...
>  > where is that special code?
> 
> The special code is the addition of the
> DSDB_CONTROL_CURRENT_PARTITION_OID in
> replmd_extended_replicated_objects() and the associated code in the
> partition module to handle the case of the upstream modules already
> having set that control.
> 
> What happens is that both the base partition and the leaf partitions
> each get a record for the partition mount points. So during the
> initial DRS replication we got an "already exists" error for the add
> of the base of each leaf partition as the search code found the
> object in the base partition.
> 
> The solution was to use the above control to tell the partition module
> to direct the add to the specific partition that the replication was
> happening on. 

I think the repl_meta_data module should not pass the control down to
the partition module when applying replicated changes. Then the
partition module can find out the correct location of the object.
And the repl_meta_data module should search the existing object by guid
in all partitions.

> One side effect of this is that if you get the GUID of the root of
> each partition and then do a search for objectGUID=theguid with
> search_options=2 so that you do a whole of tree search, then you get
> back two hits! Andrew has created a bugzilla entry for that.

Good! Maybe we also need to change the vampire code to store
CN=Configuration not in the DC=DOMAIN partition.

metze

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 252 bytes
Desc: OpenPGP digital signature
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20090902/97d569e3/attachment.pgp>