Relays do restart for different reasons. For example, two of my relays (the upgrade process is automated on both) were restarted on March 23rd, practically at the same time.
Different OS versions on different machines from different ISP for both relays.
One was restarted because some libraries needed to be upgraded on the system:
The following packages will be upgraded:
libbsd0 libc-bin libc6 libxslt1.1 locales
5 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
[...]
Restarting services...
[...] tor@default.service [...]
The server even needed to be rebooted afterward.
The other one had a very different reason, as you can see from the syslog:
Mar 23 03:07:43 leslie kernel: [849147.801728] Out of memory: Killed process 930 (tor) tot al-vm:850180kB, anon-rss:620800kB, file-rss:0kB, shmem-rss:0kB, UID:111 pgtables:1652kB oom_score_adj:0
Mar 23 03:07:43 leslie systemd[1]: tor@Leslie.service: A process of this unit has been killed by the OOM killer.
Mar 23 03:07:44 leslie systemd[1]: tor@Leslie.service: Main process exited, code=killed, status=9/KILL
Mar 23 03:07:44 leslie systemd[1]: tor@Leslie.service: Failed with result 'oom-kill'.
Mar 23 03:07:44 leslie systemd[1]: tor@Leslie.service: Consumed 2d 5h 54min 39.122s CPU time.
Mar 23 03:07:44 leslie systemd[1]: tor@Leslie.service: Scheduled restart job, restart counter is at 1.
Mar 23 03:07:44 leslie systemd[1]: Stopped Anonymizi ng overlay network for TCP (instance Leslie).
Mar 23 03:07:44 leslie systemd[1]: tor@Leslie.service: Consumed 2d 5h 54min 39.122s CPU time.
Mar 23 03:07:44 leslie systemd[1]: Starting Anonymizing overlay network for TCP (instance Leslie)...
That was the last time both relays were restarted (~ 3 days ago)
Denny
On 03/26/2024 04:54 AM Alessandro Greco via tor-relays