I’ve done the math for how long it’d take to randomly guess the last several kilobytes until something checksummed correctly.
I was not pleased with the answer.
wait until you hear about collisions (missing more bits than your hash output length guarantees a collision on average)
That would put those crypto miners to better use at least
If its a piece at the beggining or end of the file it would likely be significantly easier
On average it’ll take 2^(n-1) guesses to reconstruct 2^n bits, so… depends on how many hashes / sec you can do.
Let me save you some time: not enough.
Haha agreed, if we’re talking about kilobytes of missing data brute forcing is intractable.
There may be structure to exploit in the data format. E.g. if you’re recovering missing content from a book written in English, you can probably get away with enumerating only printable ASCII and 90% of the letters will be lowercase.
But practically, I am unconvinced because the information density is pretty high on the kinds of things people like to torrent.
You know I never thought of that… but yeah that would be a good very very very very large number.
Like throwing puzzle pieces in the air and getting it to land completed.
Could be me seeding the entire torrent except the readme file
Lmfao. Hilarious.
The annoying thing about this is that automated systems (like Radarr/Sonarr) will never get notified that it’s complete, then.
If you go to your torrent client and disable the missing file, it should get reported as “complete” to the *arrs. Manual and annoying, but it works.
thanks
Easy. Next time just download the last percentage first.
Force Recheck.
I have a lot of these just go to 100% after checking the downloaded files.
I call them “zombie torrents”
What does the 0.1% of the file contain anyway, if it’s a video and most of the data is there it might be either playable or if not it probably might be able to be repairable so it can play, albeit with minor corruption in the damaged part.
Definitely had some luck with this with videos. Hit or miss, but worth a go
Video files can be played with as little as 5% downloaded, so long as the header and footer are complete
very loose definition of “played”
It works well with shows you want to start watching immediately. Enable sequential download, 2 minues to grab the header/footer, and you can start watching. It’ll download faster than you watch.
well i know that… it’s basically streaming from the torrent…
Can those be reconstructed if necessary?
BitTorrent has partial seeding. So if someone extends a torrent with some files, the original one can still be used for seeding.
Another reason for the last bit being the slowest is because populars chunks are downloaded first.
If it’s a video, you can probably still watch it
It’s usually just an nfo or srt file that one seeder has deleted
One time I added a subtitle file to the download folder, renamed to the same as the movie file, and the download percentage jumped to completed
I wish people would adopt torrent V2 because that one missing 500 byte file can make the video unwatchable. With V2 each file has it’s own sha256 hash and can be checked and shared individually. It would also improve torrent health.
The next best alternative would be BiglyBT’s Swarm Merging feature (which works similarly, and amazingly well on v1 torrents considering it only stores a precise file size instead of a hash in Vuze/Bigly’s own DHT). I’ve been able to ‘complete’ numerous separate torrents where availability was <1.
BiglyBT already supports v2 but dunno if Swarm Merging works with such torrents yet.
Interesting. Not sure if qBittorrent supports that, I really hate switching clients :D Swarm merging for V2 should be implicit because each file has a unique hash code. So you can’t not merge.
Another thing torrent clients could do: Every torrent that is downloaded and “rechecked” automatically generates and “upgrades” a V1 torrent into a V1/V2 hybrid torrent for sharing. And when you add a normal magnet link you could get the hybrid v1/v2 torrent from others via DHT. So theoretically only one person needs to generate this upgraded torrent and it’s not up to the uploader / tracker.
Torrent V2 allows the creator to change the files in the torrent. They can replace good files with bad files etc. It’s not a perfect solution.
No that is not true. I’m not sure why that silly notion was spread, I’ve seen it on reddit too. Theoretically the protocol does allows for an extension for this but it’s not implemented and would need special considerations to do. And any client implementing this would not just swap files willy nilly, they’d implement some kind of permission or opt in. There are potential applications for this but not for regular torrents.
Yarr you need some wind in yer sails matey
What is it? If it’s a video it’s probably fine.
No it’s not. You try wating for an hour with your dick in your hand edging while waiting to see some serious… ahem… debates over global warming.
what
What, you’ve never edged while waiting to see a heated debate?
I edge the whole time and only release once the post debate analysis begins.
Huh?
Porn. He’s talking about porn.
Yeah but that is video, if you have 99.9% of the file often times it plays just fine.
You should get your money back!
Don’t worry, some hero without a cape will appear for you and seed that bitch! (wait, that sound better in my head).
Sounds fine to me.
Phew, thanks.
I recently had a torrent finish after a year and a half. It wasn’t something I was really concerned about but it gave me a nice feeling to know it had finished.
What in tarnation? who did that to you, what monster…
what in the world compelled you to want it that way?
IDK. It was hung up at 70 something % and I never got around to deleting it. Then one day after having long forgotten about it the notification saying it was done popped up and I was like “holy shit!”
Once or twice I’ve gone and found another source for the download, copied it into my torrents folder, forced my torrent client to re-scan the file and started seeding it.
Watching a thousand other clients tick over from 99% to ‘seeding’ is weirdly gratifying.
Not all hero’s wear capes. Also just so you know there’s a script which does this automatically
Details?
That’s nice, where?
Check the files included in the torrent. Sometimes the folders include a little readme or something that people set to not download.
Yes I’ve even had a video file stop at 99.9% and it still played just fine.
For video files I always set it to download first and last parts of files first. You can watch a video fairly well with like 50% downloaded if the file has the first and last section, which contain the data about how the video is stored. It’ll have occasional glitches, but it mostly works. At 99% it’s effectively all there and you may not even notice that last 1%, let alone 0.1%.
Why do people do this? Readmes and nfo files take up literal kilobytes… even over hundreds or even thousands of downloads, at most it’s going to take up a few extra megabytes of download/storage, they’re not saving anything at all. And it can be nice when the nfo includes all the releaser’s original encode settings and stuff.
¯\_(ツ)_/¯
Why do I need to know that “Torrent was downloaded from TheSiteINeverVisited.nfo”? It’s just extra noise I don’t need.
Extra files are messy.
true, but my jellyfin server doesn’t care. if i didn’t want it, i would’ve unselected it when importing the torrent.
Wouldn’t that still leave the user waiting unless they too unselected it?
Yea sometimes I’ll exclude the .nfo from my downloads. Thankfully the tracker I’m on now disallows any files that aren’t media in their uploads.
But nfos are useful
And can contain some fun ascii art.
And hateful monologs.
We love those especially.
Part of why I moved to usenet.
Everything always downloads at full speed (limited by disc write speed in my case), so if there’s missing data you find out about it within a min or two instead of after 3 days of trying.
Usenet also includes parity data so you can rebuild missing data to an extent.
I use both. A lot of stuff just isn’t on usenet.
Till you’re missing that one article.
Yup, point is I find out much much sooner and can move on to a new nzb. A single ~15gb nzb takes 5min max whether it succeeds or not. I’m never ever waiting on slow seeds.
Multiple providers can improve availability, but I’ve seen no need. Everything myself or my users have requested has been found and downloaded within 25min, including re-tries. Typically it’s about 15min from user request to ‘available to watch’ email notification.
Worse case I can fallback to torrents, but I haven’t had to yet with close to 15tb out of usenet alone.
I’m on private trackers and we don’t have problems either.
Private trackers can be a bitch to get into, and you have to re-seed what you download exposing yourself to copyright claimants and/or pay for a vpn on top.
I just raw dog a usenet server for 5min/movie and I’m done. Faster, easier, and risk free.
I’ve yet to find a modern use for usenet as I’m not in the habit of downloading everything as it comes out, nor of looking for content within a few days of release. Often I’m looking for 2-5 year old content or back catalog, and usenet has been a uniform landscape of incompletes, even with two blocks on independent providers (or they were when I bought the data blocks).
The requests I receive from my users are typically from the 1990s up to current releases, but have gone as far back as 1951.
I’ve only failed to find 5 of 140 requests since May.
