Eric Cheng via network-health:

Hi all,

I checked a few webtunnel bridges, and their "First seen" dates are all showing "1970-01-01 00:00:00". https://metrics.torproject.org/rs.html#details/7A78527DBDB7EFF85E5AA926A1B92... https://metrics.torproject.org/rs.html#details/6996BC2264025E9A87D51324CF56B... https://metrics.torproject.org/rs.html#details/81895A2C4E22E861615DD91275439...

Right now the bridges show some different and potentially reasonable dates. I am not sure what happened in the case of those bridges but we do have cases where firs_seen dates get reset to the epoch. See: https://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/40...

For the top relays, I checked a few, and instead of showing the year 1970, they showed dates with times exactly at certain hours. I feel like this kind of coincidence is weird.

2024-08-25 00:00:00 https://metrics.torproject.org/rs.html#details/9AB93B5422149E5DFF4BE6A3814E2... 2025-06-26 21:00:00 https://metrics.torproject.org/rs.html#details/5835631C79E55CDFDCF9E2D14CB99... 2019-02-18 00:00:00 https://metrics.torproject.org/rs.html#details/9AB93B5422149E5DFF4BE6A3814E2...

Is this correct behavior?

For the relays, it seems so for the bridges re-setting the first seen date to the epoch is certainly not. Georg

Information for relays was published: 2026-02-27 05:00:00 UTC. Information for bridges was published: 2026-02-27 03:57:24 UTC.

Thanks, Eric

_______________________________________________ network-health mailing list -- network-health@lists.torproject.org To unsubscribe send an email to network-health-leave@lists.torproject.org

Hello, This issue occurs in Onionoo when there are problems with Collector, as has been the case over the past few weeks. In our current pipeline, Onionoo fetches several documents from Collector. These documents are then processed and used to provide the information exposed through the Onionoo API and relay-search. If there is any misalignment between Collector and Onionoo, some of the data may not be set correctly during processing. One of the fields that is particularly affected in this situation is the |*first_seen*|field. We are currently working on a new pipeline that should address these kinds of issues among other things. However, the API for this new pipeline is not currently online, we took it offline during a recent database migration, so I cannot provide a solution at the moment. If you are a researcher and you are interested some particular data points that is messed up on onionoo maybe I could help you with that. Talk soon, -hiro On 6/3/26 03:43, mossy_road_523--- via network-health wrote:

I used the Advanced Search box on Tor Metrics to look for all running bridges with a start day between 9999 and 30000 days ago (the epoch was 20556 days ago). The webpage complained about the size of the result set, and handed back the first 2000. So, there are a lot of bridges in the "First Seen == epoch" state. (The tool wouldn't allow a starting value > 9999, so I had to use that. 9999 days is > 27 years, so there's certainly no bridge with a real First Seen date from over 27 years ago.) _______________________________________________ network-health mailing list --network-health@lists.torproject.org To unsubscribe send an email tonetwork-health-leave@lists.torproject.org