We’re in the process of moving DHCP service to our IPA LDAP server. IN our environment it
makes sense to include DHCP as part of our centralized system management scheme, which is
based on IPA. We seem to be getting about a DHCP request per second, so I don’t see this
causing a performance problem.
As part of this I’ve created a plugin file that defines most of the DHCP commands (global
config, subnets, hosts, groups, and pools — not IPV6, classes or subclasses, but they
wouldn’t be hard to add following the examples in the file). Because adding a subnet
requires restarting the server, I’m concerned about changes in LDAP having errors that
would prevent a server start. For that reason, the plugin writes a file with the object
whose configuration is changing, and calls dhcpd in test mode to verify that the
configuration is OK. If not, the error information is returned to the user. This isn’t
necessary for host entries, as they are read directly from LDAP. So changes that have to
be checked are comparatively rare.
We don’t currently using dynamic address allocation, so we don’t have to worry about
coordinating leases. If we did, we’d using the DHCP server’s standard mechanism, and
wouldn’t try to put leases into LDAP.
Show replies by date