Due to unfortunate circumstances (multiple moves and multiple machine failures), I have had the ability to do some investigating of the 'Ignored peer' bug/error across 2 Windows 7 machines, and one Win 8.1 machine that I later upgraded to 7. Additionally, during this testing, I used 4 different internet connections from 2 different ISPs, TWC and Comcast.

Aside: I don't know why people aren't putting more ass behind getting this problem/bug resolved/fixed. It's significance is minimized when dling recent torrents, but can completely put a stop to things when redownloading older torrents. Some of my downloads are stalled at 8% because all 8 seeders are ignored, and, if I un-ignore, I can't even get a complete piece from them before they are re-ignored.

My notes and research:

1) I run all of my machines without AV. There is no AV to disable, as a fix from an earlier thread suggests. Most non-crucial windows services are also disabled (e.g. Windows Search). This should minimize the ignore problem if it were related to software interference (which seems to be the case for some posters, but not me). This problem seems to be independent of the environment it is run in.

2) The problem seems to be independent of peer location. Of a sample of 38 ignored peers (from ONE torrent!!), the average peers-per-country are slightly less than two. America had the most, at 5, followed by 4 in the UK.

3) The problem seems to be independent of peer client. Of a larger sample of ignored peers, the most common client was uTorrent 3.4.2, but many other clients were represented (libTorrent 0.13.3, BitTorrent 7.x, Transmission 2.82, Vuse 5.4.0.0, Ares, 2.x.x etc). None of the ignored peers claimed to use Tixati, but this may be a function either of Tixati's small adoption, or of spoofing as uTorrent. None of the other peers sampled claimed to use Tixati at all.

3) The error seems to be independent of whether or not the peer sent blocks. Many peers hadn't sent any pieces at all before they were ignored. A little over half of the ignored peers HAD sent data with some sort of error attached. Errors, in order of prevalence, for those with errors:
->Received block out of sequence
->Incoming block not in request stack
->No piece for incoming block

4) The error seems to be intermittent. That is, when I would do a mass unignore to check the logs when set to logging level 4, suddenly everything behaved perfectly (though many of the ignored peers were offline).


I fully expect the error to return tomorrow, so I will update this thread with more digested data as soon as I can collect it next.

Hopefully this is a start. Let's lay this one to rest once and for all,
-S