[tor-relays] jumping clock

[Justin Aplin]

On Aug 8, 2011, at 3:59 PM, Softail wrote:

> Every few days I find a log message like this
> 
> Aug 08 07:20:09.241 [warn] Your system clock just jumped 122 seconds
> forward; assuming established circuits no longer work.
> 
> Seems to always be 122 seconds as far as I can tell and happens at
> random times. Doesn't seem to be correlated with the log rotations or
> anything else. I'm running ntpd and that seems to be working fine. I
> started to let it use the kludge local clock in case the network was
> unavailable. I didn't think that was likely and it doesn't seem to have
> made any difference. Google turns up some complaints about this on much
> older versions 2.0.23. I'm currently running 2.1.30 from the Tor
> repository. It's i386 running on x86_64.

I have this exact same problem about every 12 hours, always 122 seconds. As far as I can tell my system clock is never more than 1-2 seconds off. I'm running 2.3.2-alpha on a PPC OSX 10.4 machine.

> I suppose I could try
> rebuilding a 64 bit version but so far haven't built my own yet. Would
> eventually like to dig into the sources but don't have time right now.

I built my own from scratch, with and without static libraries. Hasn't helped.

> Anybody else see this or have suggestions about what to do about it?

I'm not particularly annoyed by the log entries, but I would like to know what effect it's having on my node. Are there any consequences besides a few circuits dropping every 12 hours?

~Justin Aplin