Clock jumps on relay due to hostname lookup timeouts
The security update (Tor v0.2.8.9 running on FreeBSD with Libevent 2.0.22-stable, OpenSSL LibreSSL 2.4.3 and Zlib 1.2.8.) shows the following log entry each hour:
Oct 19 02:51:07.000 [warn] Your system clock just jumped 136 seconds forward; assuming established circuits no longer work. Oct 19 02:51:07.000 [warn] assign_to_cpuworker failed. Ignoring. ... Oct 19 02:51:07.000 [warn] assign_to_cpuworker failed. Ignoring. Oct 19 02:51:15.000 [notice] Tor has successfully opened a circuit. Looks like client functionality is working. ... Oct 19 03:51:10.000 [warn] Your system clock just jumped 138 seconds forward; assuming established circuits no longer work. Oct 19 03:51:11.000 [warn] assign_to_cpuworker failed. Ignoring. ... Oct 19 04:50:37.000 [warn] Your system clock just jumped 105 seconds forward; assuming established circuits no longer work. Oct 19 04:50:37.000 [warn] assign_to_cpuworker failed. Ignoring. ... Oct 19 05:51:14.000 [warn] Your system clock just jumped 142 seconds forward; assuming established circuits no longer work. Oct 19 05:51:15.000 [warn] assign_to_cpuworker failed. Ignoring. ...
The warning first appeared on 2.8.7 after update on September 13th (Tor v0.2.8.7 running on FreeBSD with Libevent 2.0.22-stable, OpenSSL LibreSSL 2.4.2 and Zlib 1.2.8.). That time I switched back (Tor v0.2.7.6 running on FreeBSD with Libevent 2.0.22-stable, OpenSSL LibreSSL 2.4.2 and Zlib 1.2.8.) and the warning disappeared.
Lately those warnings pop up:
2016-10-21_09-03-15_noticesJ5.log:Oct 19 13:39:55.000 [warn] assign_to_cpuworker failed. Ignoring. 2016-10-21_09-03-15_noticesJ5.log:Oct 19 13:40:00.000 [warn] circuit_mark_for_close_: Bug: Duplicate
call to circuit_mark_for_close at src/or/onion.c:182 (first at src/or/command.c:559) (on Tor 0.2.8.9 ) 2016-10-21_09-03-15_noticesJ5.log:Oct 19 14:39:55.000 [warn] Your system clock just jumped 151 seconds
forward; assuming established circuits no longer work. 2016-10-21_09-03-15_noticesJ5.log:Oct 19 14:39:55.000 [warn] assign_to_cpuworker failed. Ignoring.
2016-10-21_09-03-15_noticesJ5.log:Oct 19 22:39:51.000 [warn] assign_to_cpuworker failed. Ignoring. 2016-10-21_09-03-15_noticesJ5.log:Oct 19 22:39:56.000 [warn] circuit_mark_for_close_: Bug: Duplicate
call to circuit_mark_for_close at src/or/onion.c:182 (first at src/or/command.c:559) (on Tor 0.2.8.9 ) 2016-10-21_09-03-15_noticesJ5.log:Oct 19 22:39:56.000 [warn] circuit_mark_for_close_: Bug: Duplicate
call to circuit_mark_for_close at src/or/onion.c:182 (first at src/or/command.c:559) (on Tor 0.2.8.9 ) 2016-10-21_09-03-15_noticesJ5.log:Oct 19 22:39:56.000 [warn] circuit_mark_for_close_: Bug: Duplicate
call to circuit_mark_for_close at src/or/onion.c:182 (first at src/or/command.c:559) (on Tor 0.2.8.9 ) 2016-10-21_09-03-15_noticesJ5.log:Oct 19 23:40:02.000 [warn] Your system clock just jumped 158 seconds
forward; assuming established circuits no longer work. 2016-10-21_09-03-15_noticesJ5.log:Oct 19 23:40:02.000 [warn] assign_to_cpuworker failed. Ignoring.
Posted in tor-relays: https:// lists.torproject.org/pipermail/tor-relays/2016-October/010727.html
A similar warning is reported in tor-talk: https:// lists.torproject.org/pipermail/tor-talk/2016-October/042425.html
There seems to be no impact by LibreSSL. Here the Tor configurations of the past:
tor-2.6.9-bin libre-2.2.3-bin libevent2-bin => ok
tor-2.6.10_1-src libre-2.2.5-src libevent2-bin => ok
tor-2.7.6-src libre-2.2.5-src libevent2-bin => ok
tor-2.7.6-src libre-2.3.4-src libevent2-bin => ok
tor-2.8.7-src libre-2.4.2-src libevent2-bin => warn
tor-2.7.6_1-src libre-2.4.2-src libevent2-src => ok
tor-2.8.9-src libre-2.4.3-src libevent2-src(libre) => warn
tor-2.8.9-src openSSL-1.0.2j-src libevent2-src(libre) => warn
tor-2.8.9-src openSSL-1.0.2j-src libevent2-src(libre) => warn
tor-2.8.9-src openSSL-1.0.2j-src libevent2-src(openssl) => warn
Hourly a block of up to hundred and more warnings with mostly the identical timestamp shows up. The warning comes on 7 of my 7 relays. May-be the setup is specific on my side (same for all relays):
- Freebsd 10.1
- Tor in Jail
- Jail by RDR/NAT, RDR with PF filter rule
- No changes during the transition from 2.6.x until 2.8.7
How can the warning be avoided ?
Get circuits lost (assuming established circuits no longer work) ?
Trac:
Username: Felixix