On Fri, Nov 25, 2011 at 01:25:39AM -0500, Tim Heckman wrote: > On Thu, Nov 24, 2011 at 11:06 PM, Andrew Pollock <apoll...@debian.org>wrote: > > > > I've actually been poking at this some more, and it turns out I've made a > > grave error in assuming that by not requesting a particular option you > > don't > > get it anyway. > > > > It turns out in testing, that even not requesting the host-name option, if > > it's set in the server, the client seems to get it, so this totally blows > > up > > my grand plans of telling people who don't like the hostname setting > > behaviour to just not request the hostname, so I have to revisit > > everything. > > > > regards > > > > Andrew > > > I'm usually not a fan of the files, depending on the software, but maybe an > '/etc/default/dhclient' file would make sense here. I don't mean to > over-complicate the "dhclient-script" here, but it would allow the > administrator to configure how the hostname is set with a simply flip of a > switch, instead of a hook needing to be made as well as having to worry > about what logic to use in the script. For example in > /etc/default/dhclient: > > set_hostname="yes" # sets the hostname every time > set_hostname="no" # never sets the hostname > set_hostname="detect" # maybe "dynamic", or something. This option sets > the hostname based on what "/etc/hostname" looks like. If the file exists > and contains no data, set the hostname. If not, have the local hostname > override the one given out by DHCP. > > I'm willing to bet, however, that there may be better options available. > > For anyone who comes looking for a fix, here is a workaround I've managed > to throw together for an exit hook for dhclient that sets the hostname and > does some logic so it only sets it if needed. Unfortunately, I do not have > the file on my local system but you can download the raw version of this > pastebin to "/etc/dhcp/dhclient-exit-hooks.d/set_hostname": > > - http://pastie.org/2917845 > > That could probably use a little bit of cleaning up, but it works.
Is this targeted at 4.2.2-1 or earlier? If it's targeted at 4.2.2-1, it should be a case of a dhclient-enter-hook that redefines the set_hostname() function. Using an exit hook means the hostname is going to flip flop, which probably isn't the end of the world.
signature.asc
Description: Digital signature