z9999 Posted August 6, 2006 Report Share Posted August 6, 2006 I'm surprised no one has suggested this to me previously, but I found that I can lessen the impact of Hash Fails by reducing the number of allowed Peer connections. It doesn't stop the Hash Fails from occurring, but it reduces the growth of the queue. Instead of having 37-38 pieces queued I now show only 7 pieces queued after suffering many hash fails. I have limited connected peers to 5 and many peers have been banned and I still show 5 peers connected. This seems to reduce the number of peers in each failure where previously each time a Hash Fail occurred 4-5 peers would show their Hash count increase and several would be banned simultaneously.I don't consider this a solution to the problem, but an aid none the less. Link to comment Share on other sites More sharing options...
Switeck Posted August 6, 2006 Report Share Posted August 6, 2006 I have mentioned before that fewer peers/seeds will likely be blamed for a hash fail if you reduce connections per torrent and upload slots, but only in passing on an unrelated thread a month or more ago. Link to comment Share on other sites More sharing options...
z9999 Posted August 6, 2006 Author Report Share Posted August 6, 2006 I must have missed that post, but thanks for pointing it out. I sure wish I could pin down the source of the hash fails though. I'm certain it is not related to the seeds or peers themselves, and appears to only affect received data. My ISP has sent people to check everything out and even replaced all my equipment, but as it would happen the failures are few when ever they are here to look. Everything might run well sometimes for 10-12 hours with no hash fails at all and then they may begin to occur with any torrent I might start. About the only good thing about it is I'm being forced to learn something I would otherwise ignore. Thanks for your responses. Link to comment Share on other sites More sharing options...
Ultima Posted August 6, 2006 Report Share Posted August 6, 2006 The routers that you have used... have you ever checked whether they were in DMZ mode? Link to comment Share on other sites More sharing options...
Switeck Posted August 7, 2006 Report Share Posted August 7, 2006 Here's a thread about BitComet that mentions problems with excessive piece requests from µTorrent:http://forum.utorrent.com/viewtopic.php?id=4519&p=1 Link to comment Share on other sites More sharing options...
z9999 Posted August 7, 2006 Author Report Share Posted August 7, 2006 Yes, and not using DMZ. Also I've run both with and without the router several times for extended periods just to verify that the router doesn't create additional problems.I have just noticed that an approximatly 18 hour period of excessive hash fails has ended and torrents that were failing about 70% of the pieces are now encountering no failures at all. Keeping a close watch and limiting the number of connections I have verified that the same peers/seeds are being used as were during the hash fails. This is another reason I feel my connection path is responsible. That, and the fact that when I have a peer located close by geographically and am encountering hash fails, the nearby peers seem to not be causing them if I can eliminate all the others. I'm just having difficulty in trying to find a way to determine what the path is. My ISP and I have a difficult time communicating as we speak different languages, but they try very hard to help. They even asked for a copy of uTorrent to try out themselves. Actually they asked for a client and I suggested uTorrent due to it's simplicity, just copy and run. Link to comment Share on other sites More sharing options...
Ultima Posted August 9, 2006 Report Share Posted August 9, 2006 The problem (almost) definitely lies with your connection then. The only thing left to confirm is that your OS or network card isn't bugged in any way. Link to comment Share on other sites More sharing options...
z9999 Posted September 4, 2006 Author Report Share Posted September 4, 2006 Just to let everyone know this problem has been cleared for me.I'm on a 2-way satellite connection and found that disabling Nettgain cleared my illegitimate hash fails. Since doing so I have had a total of 7 hash fails after many GB's of DL, which is very acceptable. It appears that Nettgain would be a desirable feature to have enabled if the documents describing it are correct, but until such time I can get my ISP's attention that they have a problem with it, I will leave it disabled as I am getting very good UL and DL with very little waste. So I feel that I have completely eliminated my equipment and software but need to work on my ISP. I would like to thank all those here who have given their assistance to me in explaining the things I did not understand and feel I have learned much, but still am not an expert so I may present more questions in the future. Thanks everyone. Link to comment Share on other sites More sharing options...
Firon Posted September 4, 2006 Report Share Posted September 4, 2006 What is Nettgain? Link to comment Share on other sites More sharing options...
z9999 Posted September 4, 2006 Author Report Share Posted September 4, 2006 From the best I can determine, it takes the place of TCP/IP in communicating with the satellite and is supposed to improve the efficiency up to 500%. I have read some of the papers describing it but not really clear on how it accomplishes what it is supposed to do. If you are interested in the papers describing it let me know and I'll try and dig up the links to them again. As best I can determine, Nettgain is used with satellite access only, but I may be wrong. And if I read correctly it applies to the S/R data between the satellite terminal box and the ground station only. Link to comment Share on other sites More sharing options...
Ultima Posted September 6, 2006 Report Share Posted September 6, 2006 http://forum.utorrent.com/viewtopic.php?pid=202713#p202713 Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.