16 Apr
2026
16 Apr
'26
10:53 a.m.
On 15/4/26 22:55, ttlns via network-health wrote: > Hey Hiro, > > Thank you for the elaborate answer! > > First of all, I would like to apologize for the initial duplicate messages which got caught by review. Unfortunately, mailman3 [1] did not show any confirmation, that a thread has or will be created and redirected (302 Found) me directly to [2]. If duplicate messages from new users are a frequent event, I suppose that adding some confirmation might help. > >> you would be able to verify the number of relays running on the network by parsing consensus documents that are publicly available at collector.torproject.org > I will give that a try and post an update to this thread. Thank you for the suggestion! > >> There might be legitimate > reasons for those negative spikes. > I would suppose that there should not be any reasons for many relays to go offline at the same time. > 1. Am I correctly assuming, that a significant drop in number of relays would create a vulnerable state for the overall network? Hey there, Let me elaborate a little. The negative spike on relay counts are probably an issue with our data pipeline. This is because data is aggregated daily so reboots in those metrics should be averaged out. From what the flag is concerned if a relay updates the OS packages, either the VM reboots or the tor daemon restarts. In this case some of the flags (usually it's the hsdir one) might be lost temporarily that's why you see the flags fluctuating (or dropping and then picking up). Does this answer your doubts? Relay operators are responsible for updating their system yes. What is your concern in this case? Cheers, -hiro > >> So if you see a negative spike and you know there was an update to one > of the libraries used by little-t-tor, it might be what caused it > > 2. Am I correctly assuming, that the relay operators are responsible for updates of their own system? > 3. If (2), how can or is the integrity of those updates ensured? Whilst the underlying software is OSS, I would doubt that many relay operators review the code thoroughly. e.g. I imagine ideally, there could/should be some voting system (possibly over the directory servers?) for verifying the integrity of an update? > > [1] https://lists.torproject.org/mailman3 > [2] https://lists.torproject.org/mailman3/hyperkitty/list/network-health@lists.torproject.org/2026/4/? > > Best, > ttlns > _______________________________________________ > network-health mailing list -- network-health@lists.torproject.org > To unsubscribe send an email to network-health-leave@lists.torproject.org