Ever Increasing Peer numbers

BitComet BitComet Bug Reports
1

I have suspected this for quite a long time, but I happen to be using 1.28

There’s a documentary TV series from 1986, which I uploaded to Demonoid about 6-7 years ago. I’m gratified that it has survived on its own so long, and I recommend it to just about anybody – but I realize that most people aren’t actually going to download and watch it. (“The Day the Universe Changed” by James Burke, if you’re curious.)

After the recent convulsion at Demonoid, I noticed that nobody had picked up the series after the changeover, so I decided to seed it again myself, along with a couple of others (Nero Wolfe, Seasons 1 and 2). On all of these torrents, I notice that the number of leechers I’m not connected to, gradually but constantly increases and never decreases.

At the moment, BitComet says the peers situation is 0/1[0/949]

While I’m sure this is very flattering and all, I’m absolutely certain that there are not a thousand downloaders for this 25-year-old TV series.

Nero Wolfe Season 2 is showing 1125 unconnected leechers. The series was very good, mind, but there ain’t no thousand people trying to download it. Guaranteed.

The recent television series, “The Mentalist” is quite popular, yes. I’ve been seeding two seasons of it on Demonoid for a while. But I seriously doubt these numbers:

Season 1 0/39[14/10728]

Season 2 0/36[55/9295]

Shame I can’t connect to a few more of those ten-thousand people, eh?

It looks to me like BC has a defect where it’s continually adding and adding and adding leechers, and never removing them, so that people who rejoin the swarm are counted as new, though they’ve been here before, and are maybe getting counted four or five times. Whatever, that number of leechers just keeps growing to absurd heights.

2

Kluelos, can you post a magnet link to one of these torrents. I’ll open them in two different clients and see what results I get.

3

Let’s see if this works

The Day the Universe Changed

Nero Wolfe, Season 1

Nero Wolfe, Season 2

When I stop these for a while (a couple of hours) and restart them, I get much more reasonable numbers of peers. For example, “Universe” now looks like 0/0[0/17] and Wolfe 1 is 0/2[1/24]

4

I must say that I had the same impression on repeated occasions. It seemed more obvious (or perhaps it only manifested itself) in cases where the connected peers to me were 0.

I remember uploading a torrent on torrentbox (it was some test I was doing on other things) and I forgot it in BC. After a while it constantly showed me over 100 peers, even though I never actually got to upload a single byte of it to anyone!

Given that I was the initial uploader of the torrent, that got my attention.

How was it possible that the torrent had more than 100 peers but none was downloading from me?

My port was open and other torrents were really seeding and uploading through remote connections without any issue on other random moments around that period I was watching (therefore I had no connection or settings issue).

When I went ahead and looked into the Peers tab I saw that my own IP was listed more than once among the peers.

I just dropped the whole issue and never reported it since it is so hard to reproduce these conditions while presenting irrefutable evidence that the peers are false.

But something definitely felt fishy. I also spotted behavior similar with what kluelos reports, meaning that occasionally I can see dead torrents (as in no activity whatsoever, meaning you don’t get a single byte) but BC displays for them several hundred peers and the number seems to rather increase than decrease, while if you look in the Peers tab 3/4 of them are reported “dead”. Even though a very long time passes and logics dictate that their number should decrease, they appear to do the contrary.

I just learned to live with that situation and not trust what BC shows me in that area; if I don’t see any activity for a very long time then I know it’s safe to say there will never be any activity on that torrent no matter how many peers BC shows.

Oddly enough this seems to happen for me only for the peers number (total in the swarm) but not for the seeds one.

There is no “double-checking” possibility where you can verify that all the listed peers are real. Furthermore, BC doesn’t even report peers gotten through PEX for instance.

There is also the matter of peers gotten from more than one source and that adds more complexity.

Perhaps if the Peers tab showed both peers gotten from PEX (in addition to the info already present) and also the overall number of unique peers gotten from all the sources listed in the Peers tab (meaning if you have DHT:10 peers and tracker:20 peers but 5 are common you get a line at the bottom of the window saying Unique peers: 25 peers).

That way you could double-check the number in the Peers tab with the one in the Task List column and also look at the IP of every peer reported in the swarm.

I know that it sounds as redundant work, but it could help reveal a possible bug.

5

I’ve confirmed this, when the task was downloaded with another client, the peers listed were never more then 10, which matched the reports of the tracker. I then began seeding it with bitcomet and it soon reported 40, then 60 unconnected peers, now over 24hours later, it’s showing nearly 500 peers.

6

topic renamed

“creeping peers” may not translate into Chinese very well.

7

Oh, I dunno… 匍匐同輩 works for me… And it almost even conveys what was meant! :smiley:

(Plus, I like way the phrase sounds. B) )

Also, I’ve seen this behavior for quite some time in BitComet. I always assumed I was doing something wrong, or was misinterpreting something, or was missing some piece of the puzzle. <_<

8

Dev team confirmed this is a bug, BC will repeated calculate the peers which the trackers return.

Dev team will try to fix this bug before next stable version is released.

9

That’s great, thanks for the rapid response, Ariel!

10

And thanks to the development team too. For such a non-critical issue, they tracked it down very fast.

Thanks also to Kluelos for reporting it.