[mythtv-users] Mythbuntu-diskless - can a client survive a server reboot?

[Ronald Frazier]

On Thu, Jan 27, 2011 at 5:47 AM, Mike Perkins
<mikep at randomtraveller.org.uk> wrote:
> I think the way to manage this might be to do some extra scripting in the
> server shutdown process. Test for the existence and presence of remote
> clients and issue them a remote reboot command at that point*, with a delay
> to ensure that they don't actually attempt to restart until the server is
> back in business.

Well, the timing would actually be quite tricky, because it's not just
a matter of a delay. You have a relatively small window of opportunity
in which to take down the nfs share. Take it down too early, and the
reboot hangs. Take it down too late and the frontend has already
started booting and hangs. You'd probably have to coordinate it better
than just a delay. If the DHCP server which provides the boot info for
your frontend is also on the server you are taking down, then you'd
want to stop the DHCP server service first, then tell the frontend to
reboot, then wait for the nfs connections to close, and finally start
rebooting the backend. Of course, on some machines this could be a
problem. In order to do this, you need to be able to set the
frontend's BIOS to perpetually reattempt a netboot. Some system will
only try 3 times before giving up and telling you to insert a bootable
disk.

-- 
Ron