Re: (ITS#6368) Bug in deleting entry in MirrorMode

[hyc@symas.com]

pitpalme@gmail.com wrote:
> masarati@aero.polimi.it wrote:
> 
>> pitpalme@gmail.com wrote:
>>
>>> I do see attributes reflecting state after "modify" (which changes
>>> "sn" attribute).
>>> Where exactly should I expect to see a note about "glue state"?
>>> Is output from "-d Any" helpful to figure what's going on (or wrong)?
>>
>> My understanding of your previous message was that you got a positive
>> result from delete, but an "Already exists" from a subsequent add.
> 
> That=92s korrekt.
> 
>> I inferred that you checked about the existence of the object and =20
>> you got
>> a negative result.
> 
> In my current test scenario I do not check.
> I assume a non-error on =84delete=93 in fact =84deletes=93, so I just =
> check if =20
> response
> code success (in Perl: '$res->code =3D=3D Net::LDAP::LDAP_SUCCESS=91).
> 
>> Apparently, my guess was wrong.  In that case, you probably hit
>> ITS#6097 <http://www.openldap.org/its?findid=3D6097>.
> 
> *hmmm* Good idea, thanks.
> But after reading ITS description I don=92t think it=92s  what troubles =
> me.
> We don=92t actually neither have =84multi master mode=93, but =84mirror =
> mode=93 =20
> and writes are done only on exactly one server, so there should not be =20=
> 
> a problem with second server producing a concurrent write/modify.
> I=92d expect the delete to be correctly propagated, but I can=92t check =
> it =20
> is is, simply because I have to little knowledge to read debug log =20
> correctly.
> =97

Paste the debug log somewhere we can see it then. Might as well use debug
level -1 while executing this sequence. And provide the debug output from both
servers.

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