[Battlemesh] time sychronization in wibed

[Benjamin Henrion]

On Sat, Aug 8, 2015 at 1:41 PM, Daniel Golle <daniel at makrotopia.org> wrote:
> Dear testbed team,
>
> I just stumbled into the fact that GNUnet needs at least roughly
> synchronized time for things to work.
>
> https://gnunet.org/bugs/view.php?id=3932
>
> Obviously, running systems without an RTC in an isolated environment
> will not provide that luxury to software running and the only lesson
> I still remembered out of all the things I should have learned in the
> Distributed Systems course I once visited is that one should never
> rely on the "wall clock" aka. "planetary time" in distributed systems.
> However, GNUnet partially violates that rule and doesn't provide any
> internal means to synchronize clocks (yet).
>
> Thus it would be nice to have at least rdate or NTP on all nodes on
> the testbed and use it to synchronize time using the management mesh
> and touch a file inside the overlay which can then be used after reboot
> to at least get a rough idea of absolute time from the atime/mtime of
> that file which should be good enough for GNUnet to work.
>
> Any ideas?

That remembers me the debugging I spent on curvetun vpn to find out it
does not work if the 2 nodes are not in sync:

https://github.com/netsniff-ng/netsniff-ng/issues/141

I think the crypto is very sensitive to time differences.

You might need an external clock source, like a GPS receiver if you do
not have network.

-- 
Benjamin Henrion <bhenrion at ffii.org>
FFII Brussels - +32-484-566109 - +32-2-4148403
"In July 2005, after several failed attempts to legalise software
patents in Europe, the patent establishment changed its strategy.
Instead of explicitly seeking to sanction the patentability of
software, they are now seeking to create a central European patent
court, which would establish and enforce patentability rules in their
favor, without any possibility of correction by competing courts or
democratically elected legislators."