Wake on LAN, machine on different subnet

[Colin Law]

On Mon, 24 Aug 2020 at 10:36, Olivier <Olivier.Nicole at cs.ait.ac.th> wrote:
>
> Colin Law <clanlaw at gmail.com> writes:
>
> > I am trying to use WOL to wake up a machine on a different subnet of
> > my local network.  The subnet is accessible from the machine running
> > the wakeup command.  I realise that WOL uses UDP, which will not
> > normally cross the subnet boundary.
>
> Why do you assume UDP would not cross the subnet boundaries?

My understanding may well be lacking in this area.  In
https://en.wikipedia.org/wiki/Wake-on-LAN#Subnet_directed_broadcasts
it says "A principal limitation of standard broadcast wake-on-LAN is
that broadcast packets are generally not routed. This prevents the
technique being used in larger networks or over the Internet."  I
assumed that the -i option in wakeonlan somehow got over this problem.
I may well be mistaken.

>
> You need to clarify the picture, because my sugestion would be: "if it
> works without using the option -i, don't use the option -i". But it is
> obviously not the right answer, so the question is not clear enough.
>
> It appears to be a routing problem so you need to explain what are your
> subnets, how they connect to eachother (NAT?) and all that jazz.

It works without the -i only if I run it from another machine on that
subnet, normally there is no other machine on that subnet.  This is
the organisation:

Internet
|
Router 1
|_ Machine 1 (to be woken) on subnet 10.x
|_ Router 2
    |_ Machine 2 (the one triggering the wake) on subnet 73,x
    |_ ...

I have this arrangement so that the machine 1 (which is open to the
internet) cannot access the 73.x subnet machines, but those machines
can access machine 1, and the internet.  Router 2 is configured with
Router 1 as its gateway.
Perhaps this is just a matter of correctly configuring Router 2 to
route the WOL package.  It is running OpenWRT so it may well be
possible.  I have done some googling but not come up with anything
particularly helpful.

This is all because Machine 1 (A Lenovo X201 laptop) does not seem to
have a BIOS option to auto-boot on power connection, so if the power
fails long enough for the battery to run down it will not restart when
the power is restored.

Thanks for helping

Colin