See bug #150582 in Red Hat Bugzilla (URL filled). The nmbd dies right after few logins to Samba PDC on Red Hat Enterprise Linux 4 (samba-3.0.10-1.4E from initial RHEL4 release with glibc-2.3.4-2). The nmbd log contains: *** glibc detected *** free(): invalid next size (fast): 0x08bfc970 *** The same configuration works under RHEL3 (samba-3.0.9-1.3E.2), but RHEL3 has older glibc (glibc-2.3.2-95.30) which is unable to detect this sort of bugs. See bug #150647 in RH Bugzilla and Release notes for RHEL4 for more info about new glibc features http://www.redhat.com/docs/manuals/enterprise/RHEL-4-Manual/release-notes/as-x86/ The machine with Samba is Dell with P4 CPU, no HW issuses so far. The server works well after downgrading to RHEL3. I set critical severity as this could be possible security bug. I tryed to compile latest 3.0.11 Samba with no luck. I did not tested older version of Samba on those RHEL4 machine.
Same problem here. RHEL4 with samba-3.0.10-1.4E.2 Attaching nmbd.log
Created attachment 1566 [details] nmbd.log contains multiple nmbd deaths including some normal shutdowns
3.0.10 is really old from our perspective. Unless you can (a) reproduce this against 3.0.20b, or (b) get a full backtrace with debugging symbols, or (c) get a valigrind log, it will be extremely hard to track down. If we have already fixed the bug in a later release, you will need to contact RedHat to get a patch for your platform.
Confirmed that on RHEL4, using current RPM from www.samba.org does seem to fix the problem. I do have a valgrind log, however... can be found at https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=172713
Adam says it's fixed. Closing. Volker