Hi David, This issue should have been solved and I think was caused by a bug in computing the end of the month tarball in collector. Let me know if this is not the case. Cheers, -hiro On 4/5/26 16:06, David Fifield via network-health wrote:
It used to be that the "archive" directory of collector.torproject.org would serve unchanging descriptor tarballs from past months, along with a more or less continuously updated tarball for the current month. But now it seems that there is no tarball for the current month. See here:
https://collector.torproject.org/archive/bridge-descriptors/extra-infos/ https://web.archive.org/web/20260504132617/https://collector.torproject.org/...
The date is currently 2026-05-04, but the latest tarball is
extra-infos-2026-04.tar.xz 2026-05-02 03:42 218M
Is there a way to get, for example, descriptors from May 3, without waiting until the end of the month?
There is the "recent" directory which might work, but it only goes back a few days. At this moment, "recent" covers from 2026-04-30 to 2026-05-04, but if I were to check on 2026-05-20, say, I might not be able to get descriptors from 2026-05-10.
https://collector.torproject.org/recent/bridge-descriptors/extra-infos/ https://web.archive.org/web/20260504132701/https://collector.torproject.org/...
The reason I'm asking: When I make monthly graphs of snowflake bridges, I usually wait a few days into the next month and download the next month's descriptor to include it with the data. The reason is that the tarball for May, for example, contains descriptors that cover the end of April. Without including the beginning of the next month's data, I found, the last 1–2 days of the previous month's data can be incomplete.
For example, I just used https://gitlab.torproject.org/dcf/snowflake-graphs to compute statistics using tarballs up to extra-infos-2026-04.tar.xz. This is showing just one bridge (which consists of 12 tor instances). Notice the last few days of April 2026, the "coverage" drops from from 12.0/12.0 to 9.95/12.0:
date, fingerprint, transport, users, num_instances,coverage 2026-04-24,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,14592.87,12.00,12.00 2026-04-25,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,14615.31,12.00,12.00 2026-04-26,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,14992.47,12.00,12.00 2026-04-27,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,15056.68,12.00,12.00 2026-04-28,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,15250.09,12.00,12.00 2026-04-29,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,15034.07,12.00,11.92 2026-04-30,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,12877.20,12.00,9.95
If I regenerate the statistics, manually including also the "recent" descriptors, the coverage gets "filled in" (and the users count correspondingly goes up):
date, fingerprint, transport, users, num_instances,coverage 2026-04-24,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,14592.87,12.00,12.00 2026-04-25,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,14615.31,12.00,12.00 2026-04-26,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,14992.47,12.00,12.00 2026-04-27,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,15056.68,12.00,12.00 2026-04-28,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,15250.09,12.00,12.00 2026-04-29,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,15099.58,12.00,12.00 2026-04-30,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,14834.64,12.00,12.00 2026-05-01,5481936581E23D2D178105D44DB6915AB06BFB7F,snowflake,14781.53,12.00,12.00 _______________________________________________ network-health mailing list -- network-health@lists.torproject.org To unsubscribe send an email to network-health-leave@lists.torproject.org