(ITS#3988) Slapd crashing randomly after a failed bind

[joachim_nock@idg.de]

Full_Name: Joachim Nock
Version: 2.2.28
OS: SuSe Enterprise 8
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (213.61.162.210)


Hi,

I'm having trouble with a OpenLDAP server, v2.2.28 running on SuSe Enterprise 8,
using the meta module to pass data through from another LDAP server. However,
when IBM WebSphere utilizes this proxy, slapd crashes after a while. V2.2.27 had
the same behaviour, and showed that it doesn't even happen at the same line of
code, but that there are different possibilites. One example in a log
(unfortunately not at loglevel -1) would be

conn=627 op=13 SRCH base="cn=was5test,ou=techuser,o=idgaetest,c=de" scope=0
deref=3 filter="(objectClass=*)"
conn=627 op=13 SRCH attr=1.1
request 4 done
=>meta_back_search [0] match="" err=""
==>meta_back_dobind: (anonymous) bind failed with error -1 (Can't contact LDAP
server)
==>meta_back_dobind: (anonymous) bind failed with error -1 (Can't contact LDAP
server)
conn=627 op=11 SEARCH RESULT tag=101 err=0 nentries=1 text=
slapd: io.c:171: ber_free_buf: Assertion `((ber)->ber_opts.lbo_valid==0x2)'
failed.
conn=627 op=10 SEARCH RESULT tag=101 err=52 nentries=0 text=

The only common thing is that meta_back_dobind fails each time just before it
crashes - however, if it doesn't always crash when meta_back_dobind is failing.
;)

What would be interesting in the first place would be a way to gather more
important data, at least the log at level -1 seemed not too much more
interesting (to me at least ;) ), and I can't reproduce the bug using
ldapsearch.

Thanks in advance,
Joachim Nock