How tor bridge work
We are omitting the packet captures from our volunteer for privacy reasons, but they did provide us some insight on the blocking mechanism used.
Almost all of the relays that were not in the earlier consensus were blocked. The reachability data of Tor relays suggests to us that the censor is automatically updating its block list based on the Tor consensus. Of these emergent relays, all but one timed out when a direction connection attempt was made.
These relays may have been new, or they may have been existing relays that were just not online at the time of our first measurement on February 25th. There were a total of 952 relays that were present in the March 3rd consensus but not present in the February 25th consensus. On March 3rd, 6966 out of 7064 relays in the consensus timed out.Īs expected, there was some churn as new relays joined the network and old relays dropped off. Of the 6647 relays in the public Tor consensus on February 25th, 6647 of connection attempts to these relays from the A1 ISP in Belarus timed out with an error consistent with the blocking pattern described below. A volunteer ran reachability scans of the entire Tor consensus from within the A1 ISP’s space on two different days, February 25th and March 3rd, to measure how agile the censor was at updating their Tor relays block list. We noticed an attempt to enumerate and block access to all Tor relays and kept their block list relatively up to date with the current consensus. Here we saw the number of directly connecting users plummet and again a rise in bridge users, that continues to the time of writing this report.īlocking the Tor directory authorities and the fallback directories will effectively prevent new users (or users with an expired guard) from downloading the Tor consensus, but so will the complete blocking of all Tor relays. We have data for the following ASes in Belarus, listed here along with their organization name (according to the latest whois data): In most places, censorship is not consistent across different ISPs or autonomous systems. Starting in February of 2021, we saw increasingly sophisticated enumeration and blocks of Tor’s bridges and bridge distribution infrastructure. This consisted of blocking access to Tor relays and directory authorities. Then, in October of 2020, Belarusian censors undertook a renewed effort to block access to Tor. This initial period of blocking was short-lived and connectivity to the Tor network was returned a few days later. We first detected blocking of the Tor network in Belarus in August of 2020, occurring alongside widespread Internet shutdowns. GitLab Tor blocking events in Belarus from 2020-2021 Date: 21 April 2021įollowing the presidential election in 2020 and amid ongoing protests, Belarus has been ramping up their Internet censorship efforts by using increasingly sophisticated methods to block access to anti-censorship tools.The Tor Project / Anti-censorship / censorship-analysis.Link to original report: reports/2020/belarus/2020-belarus-report.md Reposing here some investigation we did into the blocking of Tor in Belarus.