named contains high number of assertions checking expected state of the
daemon. That is part of defensive code style to prevent many attacks.
The most common failure is failing some assertion check in rare
circumstances. Even when this should not happen, try keeping the service
running. If such failed assertion produces coredump just from time to
time, avoid failing hard the whole service. coredumpctl will keep track
of all crashes anyway.
Return failed status code to command. Not only report error message to
the log, but also report reload success. Must not terminate running
service on failed reload.
BIND is able to react to network configuration changes and therefore it can start even before all interfaces are fully configured. There is no need to wait until interfaces are fully configured.
Signed-off-by: Tomas Hozza <thozza@redhat.com>
- Split chroot package for named and named-sdb
- Extract setting-up/destroying of chroot to a separate systemd service (#997030)
Signed-off-by: Tomas Hozza <thozza@redhat.com>
- don't build ODBC and Berkeley DB DLZ drivers
- end of bind-chroot-admin script, copy config files to chroot manually
- /proc doesn't have to be mounted to chroot
- temporary use libbind from 9.5 series, noone has been released for 9.6
yet
- removed all deprecated stuff from CVS
