Re: hpropd propagation to slave with ldap backend?

[Love Hörnquist Åstrand <lha@kth.se>]

8 nov 2006 kl. 02.49 skrev Václav Hůla:

> Andrew Bartlett napsal(a):
>> On Fri, 2006-10-27 at 11:49 +0200, VĂĄclav HĹŻla wrote:
>>> Can anyone tell me what i am doing wrong or now to stop hprop  
>>> messing
>>> with nonexistent dc=natur,dc=cuni,dc=cz~?
>>
>> Looks like a bug to me.  That code presumes that the name is a  
>> filename,
>> and that a rename is possible.  Try this (untested) patch.
>>
>> The harder part is fixing this properly.  We should probably 'create'
>> the database, then see if the db->hdb_rename element is present,  
>> and if
>> so, reopen with the ~ name.
>>
>> Andrew Bartlett
>
> I applied patch against gentoos heimdal-0.7.2-r3 (with small  
> difference
> that int optidx is int optind there), applies well after that and  
> hprop
> is working well now.

The problem with using hprop in the ldap setting is that hprop assumes
it can throw way the complete database on the recvive end. That
is not true for LDAP, as an initial load its fine, but doing propagating
is better done using the internal propagation builtin to your ldap
implementation.

I think this is a bug too, but its not possible to fix right now (of  
ever)
and its better to just tell the user its not possible to do this.

Love