[tor-relays] What could cause a huge clock skew (9 days) across Tor restarts - anyone else experienced something like this?

s7r s7r at sky-ip.org
Sat Sep 7 10:25:29 UTC 2019


Hello,

I'd like to hear if anyone else experienced this and has an idea of what
the cause might be.

I did not oped a ticket about this yet because I am not sure there is a
problem in Tor, and it eventually healed by itself.


The relay was running just fine. Server had an accurate time before
upgrade and time service ntp was running. I only did an upgrade to Tor
from the latest nightly build and of course it does a service stop ->
upgrade -> service start.

After Debian finished installing the new Tor, I did as usual: check the
log file to see that all is OK and circuits were built and descriptors
published, etc. I saw this:

The log before restart was reporting:

Sep 06 21:03:16.000 [notice] Interrupt: we have stopped accepting new
connections, and will shut down in 30 seconds. Interrupt again to exit now.
Sep 06 21:03:17.000 [notice] Delaying directory fetches: We are
hibernating or shutting down.
Sep 06 21:03:46.000 [notice] Clean shutdown finished. Exiting.

[skipped boring part intentionally]

Aug 28 07:40:07.000 [warn] Our clock is 9 days, 11 hours, 20 minutes
behind the time published in the consensus network status document
(2019-09-06 23:00:00 UTC).  Tor needs an accurate clock to work
correctly. Please check your time and date settings!
Aug 28 07:40:07.000 [warn] Received ns flavor consensus with skewed time
(CONSENSUS): It seems that our clock is behind by 9 days, 11 hours, 20
minutes, or that theirs is ahead. Tor requires an accurate clock to
work: please check your time, timezone, and date settings.
Aug 28 07:40:07.000 [warn] Problem bootstrapping. Stuck at 0%
(starting): Starting. (Clock skew -818406 in ns flavor consensus from
CONSENSUS; CLOCK_SKEW; count 2; recommendation warn; host ? at ?)
Aug 28 07:40:08.000 [warn] Our clock is 9 days, 12 hours, 20 minutes
behind the time published in the consensus network status document
(2019-09-07 00:00:00 UTC).  Tor needs an accurate clock to work
correctly. Please check your time and date settings!
Aug 28 07:40:08.000 [warn] Received microdesc flavor consensus with
skewed time (CONSENSUS): It seems that our clock is behind by 9 days, 12
hours, 20 minutes, or that theirs is ahead. Tor requires an accurate
clock to work: please check your time, timezone, and date settings.
Aug 28 07:40:08.000 [warn] Problem bootstrapping. Stuck at 0%
(starting): Starting. (Clock skew -822006 in microdesc flavor consensus
from CONSENSUS; CLOCK_SKEW; count 3; recommendation warn; host ? at ?)

and then

Sep 06 21:04:50.000 [notice] Starting with guard context "default"
...proceed to normal start and everything as usual from now...

So, Tor had the time Sep 06 21:03:46.000 before restart.

After restart, it thought it had Aug 28 07:40:07.000 and then Aug 28
07:40:08.000 and then it healed and reported Sep 06 21:04:50.000.

This is kind of odd. What could be the reason for this? The server is
just a Debian machine that runs Tor and nothing else.


Thanks!


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://lists.torproject.org/pipermail/tor-relays/attachments/20190907/c03d6b87/attachment.sig>


More information about the tor-relays mailing list