Re: segmentation fault since V2.1.16

["Pierangelo Masarati" <ando@sys-net.it>]

> A 17:05 17/04/2003 +0200, Pierangelo Masarati a écrit :
>
>>> Hi,
>>>
>>> I already posted this question, but I haven't got any response..
>>>
>>> I have sefveral LDAP serveurs, with multiple backends.
>>> All worked perfectly vith openldap 2.1.8 (and dbBerkeley 4.0.14 and
>>> openssl 0.9.6.g).
>>>
>>> I can't migrate to openLDAP 2.1.16 (nor 2.1.17) (with dbBerkeley
>>> 4.1.25 and openssl 0.9.7a).
>>>
>>> I have a segmentation fault with slapd (and slapadd, ....) with 4
>>> servers openLDAP.
>>>
>>> make test works.
>>>
>>> My configuration is like this :
>>>
>>> .....
>>> # first backend, slave, subordinate
>>> database      ldbm
>>> suffix                "ou=etudiants,dc=univ-nancy2,dc=fr"
>>> subordinate
>>> ...
>>> updatedn  "cn=adminSlave,ou=etudiants,dc=univ-nancy2,dc=fr"
>>>
>>> #second backend, slave, subordinate
>>> database      ldbm
>>> suffix                "ou=pers,dc=univ-nancy2,dc=fr"
>>> subordinate
>>> ...
>>> updatedn "cn=adminSlave,ou=pers,dc=univ-nancy2,dc=fr"
>>>
>>> # root backend, only an entry (dc=univ-nancy2,dc=fr)
>>> database      ldbm
>>> suffix                "dc=univ-nancy2,dc=fr"
>>>
>>>
>>>
>>> If I use first backend and root backend, I haven't got any
>>> segmentation fault If I use second backend and root backend, I
>>> haven't got any segmentation fault
>>>
>>> If I use first backend and root backend, and I comment updatedn
>>> (Master), I have a segmentation fault If I use second backend and
>>> root backend, and I comment updatedn (Master), I have a segmentation
>>> fault
>>
>>- Can you provide a stack backtrace?
>
> I don't know how to do this.

run slapd (compiled with debugging symbols, -g)
under gdb, e.g.

gdb <path/to>/slapd

r <your usual args> -d 0

# after crash

bt


>
>>- Did you try back-bdb instead of back-ldbm?
>
> Yes. I tried database bdb, it's the same think
>
> If I remove the two subordinate directives, It work's
> But I haven't that I want (to glue the different backends).

OK, so it should be smtg with glue.

p.

-- 
Pierangelo Masarati
mailto:pierangelo.masarati@sys-net.it