[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: How to deal with OS hibernation



I don't believe the hibernation is a problem at all. Tor messages
include a timestamp to prevent replay attacks. When you restore your
computer from hibernation, Tor picks up right where it left off, but
notes that the system clock has jumped forward. Since any messages
regarding nodes and circuits may now be invalid, Tor basically
re-bootstraps itself (correct me here if I'm overstating). This, I
think, is similar to if Tor had received a SIGHUP. Even though it is
logging a warning, since the source of the jump is known and innocuous
(hibernation), this warning should be safely ignored.

Where this warning would have more effect is if you were actively
using the network and your clock synced significantly forward (NTP).
Then you would lose your current circuits during the re-bootstrap
which would cause any connections you had over Tor to drop. Hence the
reason this is a warning level message (there may be other reasons
too).

I think you'd be fine to let Tor run through hibernation even without
your SIGTERM hibernate/restore script.
--
Marcus Griep
——
Ακακια את.ψο´, 3°

On Wed, Apr 22, 2009 at 3:56 AM, Robas, Teodor <teodor.robas@xxxxxxxxx> wrote:
> Hello everybody,
>
> I regularly use tuxonice kernel hibernation. Recently I installed tor and
> configured as
> bridge and exit point. I want to know how it is best to use tor in this
> context.
>
> Looking in the notices.log I found this warnings:
> "[warn] Your system clock just jumped 31831 seconds forward; assuming
> established
> circuits no longer work". So I wrote a small script that sends SIGTERM (15)
> when going
> into system hibernation. Tor is then restarted at system restore.
>
> What is the recommended way to deal with hibernation ? It seems hard to find
> information
> about this because of the confusion between "tor hibernation" and "system
> hibernation".
>
>
>