Red Hat Bugzilla – Bug 314821
Slapd segfaults, leaving LDAP database corrupt
Last modified: 2007-11-30 17:12:17 EST
Description of problem:
Slapd segfaults, leaving LDAP database corrupt.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Configure slapd to accept connections via LDAPS
2. Start slapd with "service ldap start."
3. Stop slapd with "service ldap stop."
4. Manually start slapd with "slapd /usr/sbin/slapd -h ' ldaps:///' -u ldap"
Slapd crashes and database is corrupt. Dmesg says:
slapd: segfault at 00000046 eip 001c5474 esp b7761800 error 4
Couldn't read floating-point and SSE registers: No such device.
#0 0x0011e3a1 in _dl_debug_state () from /lib/ld-linux.so.2
#1 0x00000003 in ?? ()
#2 0x00000006 in ?? ()
#3 0xbfe6d7c4 in ?? ()
#4 0x00000001 in ?? ()
#5 0x00000000 in ?? ()
Slapd should start.
Created attachment 213091 [details]
slapd.conf, ldap.conf and LDAP database
I am not sure if I reproduced the bug correctly, but something very strange is
happening inside Berkeley DB. It does not matter how I start slapd, but second
time I start it, it fails to read from database and crashes...
I downgraded bdb back to 4.4.20 and it works. Could you please try if it works
for you? You can find it at http://people.redhat.com/jsafrane/bugs/314821/
In the meantime I'll try to figure out what's wrong with it.
Hm, BDB 4.6 has new locking mechanism and openldap 2.3.38 does not support it
yet... Why they do not write it into release notes/readme/changelog???
Please test the version mentioned in #2 and let me know as soon as possible, I'd
like to fix it before Fedora 8 gets out.
I tried your packages (Comment #2) and they seem to work fine. I "system ldap
restart"'ed, "kill -9"'ed and rebooted. The database did not become corrupt.
Thank you very much! It's strange nobody noticed it in Fedora 8 test releases -
it seems to me nobody is using it...
Fixed in openldap-servers-2.3.38-3.fc8