Re: (ITS#5809) syncrepl + back-ldif + "rename to same DN" fails

[hyc@symas.com]

hyc@symas.com wrote:
> h.b.furuseth@usit.uio.no wrote:
>> Full_Name: Hallvard B Furuseth
>> Version: HEAD
>> OS: Linux
>> URL:
>> Submission from: (NULL) (129.240.6.233)
>> Submitted by: hallvard
>>
>>
>> With the latest back-ldif/ldif.c (rev 1.99),
>> 	./run -b ldif test048-syncrepl-multiproxy
>> breaks with "master and P1 slave databases differ" and
>> 	./run -b ldif test018-syncreplication-persist
>> when patched to not reject $BACKEND = ldif
>> breaks with "producer and consumer databases differ".
>>
>> The offender is "rename to same DN" (ITS#5319), i.e. rev 1.99.
>> Unless there is a big "duh!" in it, it exposed a syncrepl bug.
>
> Yes, thanks. The consumer assumed that an entry being renamed would have no
> other modifications associated with it. This assumption is wrong. There are no
> rename ops in test018, but apparently back-ldif is not returning the RDN with
> the same case as the original entry, so syncrepl assumes a rename (that only
> changed letter case) was done.
>
> Still looking at the fix.

This is mostly fixed in HEAD. The diff still doesn't compare because some cn 
values are in the wrong order. I think you need to check what back-ldif is 
doing that triggers the false detection of a rename op.

-- 
   -- Howard Chu
   CTO, Symas Corp.           http://www.symas.com
   Director, Highland Sun     http://highlandsun.com/hyc/
   Chief Architect, OpenLDAP  http://www.openldap.org/project/