Onionoo issueshttps://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues2024-03-26T17:06:40Zhttps://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/40044A number of bridges have weird details2024-03-26T17:06:40ZHiroA number of bridges have weird details@gk has noticed a number of bridges with weird details attributes on relay-search:
Ex: https://metrics.torproject.org/rs.html#details/A25974E6B8CB504D51D5730F6C6AFC4A5164A0D9@gk has noticed a number of bridges with weird details attributes on relay-search:
Ex: https://metrics.torproject.org/rs.html#details/A25974E6B8CB504D51D5730F6C6AFC4A5164A0D9https://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/40028Why bridges offline for a few days get first_seen reset to 1970-01-012022-09-28T07:04:34ZHiroWhy bridges offline for a few days get first_seen reset to 1970-01-01We kept this bridge out of bridge statuses for 24 hours https://metrics.torproject.org/rs.html#details/2D8C27AA9C2FEC3EC468D7ABE9C3EDCA3C86610A and the first_seen date has been reset to the epoch. We should find out why this is happening...We kept this bridge out of bridge statuses for 24 hours https://metrics.torproject.org/rs.html#details/2D8C27AA9C2FEC3EC468D7ABE9C3EDCA3C86610A and the first_seen date has been reset to the epoch. We should find out why this is happening for bridges only.HiroHirohttps://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/40025if no bridges in the network-status have Running flag, that doesn't mean they...2024-01-18T14:48:46ZRoger Dingledineif no bridges in the network-status have Running flag, that doesn't mean they're all downIn https://gitlab.torproject.org/tpo/anti-censorship/rdsys/-/issues/102 we discovered a fun surprise: since Serge, the bridge authority, uses the same approach to generate networkstatus documents as the v3 dir auths use, then if Serge is...In https://gitlab.torproject.org/tpo/anti-censorship/rdsys/-/issues/102 we discovered a fun surprise: since Serge, the bridge authority, uses the same approach to generate networkstatus documents as the v3 dir auths use, then if Serge is generating a networkstatus document within 30 minutes (TestingAuthDirTimeToLearnReachability) of starting up, it will not express an opinion about the Running flag for each bridge it lists.
The fix we're putting in for rdsys is: if *no* bridge in the networkstatus list has the Running flag, then what it means instead is: we don't have any new info about whether those bridges are Running, and we should do something smart when interpreting the data, like use the most recent info we *do* have.
I've opened this ticket in the onionoo component, but if there are other pieces of metrics that need to get smarter behavior here, please open tickets for them too. :)https://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/40023Why is onionoo discarding some descriptors2022-09-28T07:04:37ZHiroWhy is onionoo discarding some descriptorsWhile debugging https://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/40015 we found out that onionoo is discarding some descriptors that collector is instead parsing as valid.
We should investigate why this happens...While debugging https://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/40015 we found out that onionoo is discarding some descriptors that collector is instead parsing as valid.
We should investigate why this happens. I have a theory that it is because of the bandwidth and status models in onionoo itself.Metrics OKRs Q3-Q4 2022HiroHirohttps://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/40018Some relays have last_restart attribute older than the first_seen attribute2024-02-06T13:24:32ZHiroSome relays have last_restart attribute older than the first_seen attribute@gus has found a relay that has ifs 'first seen' attribute equals to 2021-12-06, but the 'last restart' to 2021-11-29.
https://metrics.torproject.org/rs.html#details/F3ED9FE52BBCB8E1662BF61346BF1FE708997EB5
I wonder if this is a bug i...@gus has found a relay that has ifs 'first seen' attribute equals to 2021-12-06, but the 'last restart' to 2021-11-29.
https://metrics.torproject.org/rs.html#details/F3ED9FE52BBCB8E1662BF61346BF1FE708997EB5
I wonder if this is a bug in how the first_seen attribute is set.
For relays see:
https://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/blob/master/src/main/java/org/torproject/metrics/onionoo/updater/NodeDetailsStatusUpdater.java#L281
For bridges see:
https://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/blob/master/src/main/java/org/torproject/metrics/onionoo/updater/NodeDetailsStatusUpdater.java#L389Metrics OKRs Q3-Q4 2022HiroHirohttps://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/40011Onionoo is i/o intensive2023-01-23T11:05:36ZHiroOnionoo is i/o intensiveOnionoo is very i/o intensive. We should find out how much an update takes to run and if we can optimize it's operations.
Maybe ultimately the solution is to move to a data store system for metrics with an API that onionoo can query so t...Onionoo is very i/o intensive. We should find out how much an update takes to run and if we can optimize it's operations.
Maybe ultimately the solution is to move to a data store system for metrics with an API that onionoo can query so that it won't need to write to files.Metrics OKRs 2021https://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/32065Cache-Control header on 404 does not permit caching2022-04-11T16:50:03ZirlCache-Control header on 404 does not permit cachingDiscovered when deploying a new frontend host.Discovered when deploying a new frontend host.https://gitlab.torproject.org/tpo/network-health/metrics/onionoo/-/issues/26132Fix the Expires header on relay search2021-10-19T14:55:41ZteorFix the Expires header on relay searchSometimes I get really stale (8 hours old) cached relay search pages. I think this happens because the page sets an Expires header in 1970. And I think my browser ignores that time, because it's before the Date header. Then it guesses th...Sometimes I get really stale (8 hours old) cached relay search pages. I think this happens because the page sets an Expires header in 1970. And I think my browser ignores that time, because it's before the Date header. Then it guesses that it can be cached for multiple hours.
Please set an accurate "Expires" header on relay search. You could set it to a few minutes after the scheduled time of the next Onionoo job. We could also add a max-age of slightly more than an hour.
Here is the relevant HTTP spec:
​https://www.w3.org/Protocols/rfc2616/rfc2616-sec13.html#sec13.2.4