On Sat, 10 Aug 2024 21:07:14 +0100 Will McDonald wmcdonald@gmail.com wrote:
It doesn't help with the lack of coredump, but does starting Bind in the foreground with increased verbosity help in terms of casting light on the root problem?
On Sat, 10 Aug 2024 at 21:02, Franta Hanzlík via users < users@lists.fedoraproject.org> wrote:
On my Fedora 40 x86_64 system named (ISC BIND daemon) crashes at start. There is active abrtd, and it after systemd-coredump action create some information about crash in /var/spool/abrt/ccpp-YYYY-mm-dd-* directory. I wanted to see some details about the crash, so I installed the debug package (# dnf install bind-debuginfo). But hey - now named still crashes on start, but abrtd doesn't create anything! There are systemd-coredump entries in the log, but not abrtd, and no errors or reasons why it didn't get to abrtd:
Aug 10 17:43:33 dc0 systemd-coredump[5806]: Process 5794 (named) of user 25 dumped core. Module wins_ldb.so from rpm samba-4.20.4-1.fc40.x86_64 Module vlv.so from rpm samba-4.20.4-1.fc40.x86_64 ... Aug 10 17:43:33 dc0 systemd[1]: systemd-coredump@6-5805-0.service: Deactivated successfully. Aug 10 17:43:33 dc0 systemd[1]: named.service: Control process exited, code=exited, status=1/FAILURE Aug 10 17:43:33 dc0 systemd[1]: named.service: Failed with result 'exit-code'.
I had an idea that some limits for systemd-coredump might need to be increased or so, but i can't find anything that solves it. Please, can anyone advise? --
Hi Will, thanks for the advice. It is not exactly a problem of bind/named, but it crashes due to Samba DLZ plugin (dlopen loaded .so library) used when Samba act as Active Directory Domain Controller. Without this plugin is bind/named working fine.