Thanks for reporting!
Currently, since yesterday we experience a technical problem with our exit nodes. These should be resolved in 1-2 days. Please, report if the problem persists after that.
Note that the problem with the exit nodes should only affect anonymous downloads. When you uncheck the “download anonymously” checkbox, the metainfo request will be immediately retried directly.
The problems described here already manifested themselves at least 2 weeks prior to yesterday though.
I can confirm LesVerres’s finding: the problem is not here since yesterday. On the other hand I can conform ichorid’s words about the workaround. Gives you seeds and peers back again. Yesterday I completely deleted every trace of Tribler, even in the registry, and after that I installed the latest stable version (7.4.4). The problem seemed to be solved by that but that only lasted for a short time.
@LesVerres, please try unchecking the “download anonymously” checkbox, or even disable it by default in the options. Getting metadata through tunnels is unreliable.
I would have expected it to work when downloading non-anonymously. Doing so would however defeat the purpose of Tribler.
Thanks for looking into it, I wasn’t sure of you were checking the forum regularly.
@LesVerres , 7.5 development have been effectively frozen during the last few weeks as the lab was busy doing other things.
Also, if you have idea how to deal with anonymous vs non-anonymous DHT checks by default, please share your thoughts.
@ichorid Considering that anonymity is Tribler’s unique selling point, I would say, DHT checks should be anonymous by default (and obfuscated if possible).
If a potential user prefers un-anonymous downloading and/or DHT checking, there are plenty of alternatives to choose from, so no need to cater towards those people.
I wouldn’t expect these DHT checks to be very demanding on the Tribler’s exit nodes, but would it otherwise be an option to relay them through the TOR network?
@LesVerres, anonymous DHT checks are more complex to implement efficiently. We got a ticket for this problem since 2017.
I agree with you. I started using Tribler a couple of years ago because it allowed to download anonymously. Telling us now that we have to disable anonymous download to make it work is like if Tor’s developers would ask us to disable all its features if we want to keep on using it… For that purpose we already have Firefox. If this issue is not solved I’m afraid most of Tribler’s users (including myself) will start migrating to other clients.
@heribator, Tribler is many things to many people. Most people use Tribler to download things non-anonymously and only use anonymity functions from time to time. The problem is how to present both anonymous and non-anonymous download options in the GUI without compromising security and usability.
@ichorid, the first you see when you enter Tribler’s website is:
“Privacy using our Tor-inspired onion routing
Search and download torrents with less worries or censorship”
I don’t know about other users, but as @LesVerres said, for me this is the main factor that makes Tribler better than other clients. Without it I would have to reconsider my choice and see if it’s still above the rest.
@ichorid I’m curious, how was it established that most people use Tribler to download things non-anonymously?
Furthermore, from the expression of this behaviour, the user’s preference to do so can’t simply be inferred (to be clear, I’m not saying that you did).
Although anecdotal, the main complaint of other Tribler “converts”, I received, was the low download speed (although it must be said, it has improved considerably). This lead to situations where users gave up anonymity in favour of download speed, even though they rather would’ve had both. This could have contributed to the behaviour you described.
I’ve read the ticket of the DHT check problem and I see the complication.
That being said, I think @heribator provided an applicable analogy, taking into account the way Tribler promotes itself.
@LesVerres, @heribator you make a valid point about users mainly coming to Tribler for anonymous downloads. However, they soon discover that our technology is far from perfect and anonymity reduces performance. Then they start to differentiate their anonymity usage.
Again, this is anecdotal. We can’t check it because we can’t track our users. However, running Tribler exit nodes gives us some insight into our users’ average behaviour. Last year we doubled our exit nodes’ bandwidth capacity, and yet this capacity is not completely utilized to this day. For me, this indicates that the default behaviour for our established userbase is to only use anonymity when it really matters. Probably for reasons described by @LesVerres
After all, presenting an option for non-anonymous download is a workaround for imperfect anonymity technology.
I must say it’s rather disappoiting the way @ichorid reacted recently. Tribler is troubled by an exit node problem, ichorid said some three days ago. Fair he did: in a complex setting problems can arouse and even persist a while, Overt pointing out the course of the problem can be effective in finding a solution. But it’s no use changing the course of the debat to wether or not users want anonymity. There are far better, a least faster, clients if your only object is to connect to seeds and peers. That needs no discussion. So back to the problem. If it was persisting after 1 or 2 days we were asked to report that. I can confirm the problem persists.
The ink of my message still being wet I see that the problem changed. Earlier this morning it fully persisted but now there’s a change: from a complete absence of seeds/peer connections I now notice a connection, not always stable but nevertheless… I will keep track of it.
It’s been three days so I just want to report that I’m still experiencing the problem. Had been for a day or two before your post, and it’s still persisting exactly the same as it was. Slightly different to what a couple others have said, my client shows both 0(0) seeds and 0(0) peers as well for all torrents with one or more hops (and instantly shoots up to higher values and starts downloading as soon as I turn anonymity off, like others have reported).
Any word on a resolution to this issue?
I’m still using 7.3.2 by the way, as any later client I’ve downloaded (7.4.0, 7.4.4, and 7.5.0) crashes on startup for me (Windows 10), but that’s a separate issue…
I’m sorry to say, but the problem is back. No seeds no peers connected (other than only for a while).
@ichorid It’s three day later, unfortunately I can’t say I see any improvements.
Last year we doubled our exit nodes’ bandwidth capacity, and yet this capacity is not completely utilized to this day. For me, this indicates that the default behaviour for our established user base is to only use anonymity when it really matters.
That seems a very bold conclusion based on this single metric. Is the exit nodes’ consumed bandwidth indicative of the actual demand of Tribler’s users (e.g. considering bugs like this one), what other factors could influence this outcome?
Furthermore, what does it tell about the absolute demand for downloads of your user base? What does it say about their considerations on whether or not to download anonymously?
If there was an overcapacity in bandwidth, was it redistributed over the current users? How did this overcapacity vary over time? Was it an average overcapacity or was it persistent? It would be interesting to see it plotted out over time.
Interesting questions @LesVerres but the main thing stays: Tribler differs from other torrent handlers and what the difference is one can easily see opening Triblers website. That leaves no doubt about Triblers identity. So the discussion about whether or not there are users that use Tribler for other reasons is distracting. You can’t say: here we have the perfect faciltity for you to become a high jumper but we don’t have bars.
Absolutely, there should be no discussion about that. In my opinion, the development aim should be to overcome the shortcomings in the program, not to facilitate users to make compromises.
Bypassing anonymity would be acceptable as a fallback measure only.