Lloyd Bryant wrote: > There is still an issue with that "Giving priority to THEX" stall. Since I > was testing against a Limewire source (which allow multiple downloads), I > tried the following: > 1. Initiate the download. Main download started, THEX download created, > but stopped with "Max number of downloads from this host" > 2. Changed "Max downloads from a single host" to 2. > 3. The THEX download started (but timed out - different problem, seems to > be a problem with the particular source I was using). > When the main download completed its first chunk, it stopped with the > "Giving priority to THEX" message, even though it *should* have been free to > continue. > So it looks like the "Giving priority to THEX" code pays no attention > whatsoever to the "Max downloads from a single host" value - if THEX hasn't > completed when the check is encountered, then the main download stops.
Yes, apparently I implicitely assumed only a single download would be allowed at a time. Initially I didn't realize that putting a download back into the queue always means dropping the current connection. Due to that this behaviour is far worse. If you update from SVN, it shouldn't stop the current download anymore with a per host limit above 1. Maybe the THEX strategy needs some improvement. Most sources are good and if you download from a single source, the THEX data is fairly useless except to guard against rather unlikely transmission errors. It would probably be better if we didn't initiate a THEX download in such cases at all. > Note: I *have* had some good luck with leaving that value set to 2 - it > usually gets the THEX download out of the way before that "first chunk" > situation hits. And I'm not all that sure that having it set to 1 is really > all the beneficial to Gnet, as I generally get the same amount of > *bandwidth* from a given source, regardless of how many downloads that > bandwidth is split between. I'd prefer a less draconian approach of sharing resources equally between peers too. There are too many limits and too many reasons to get punished. I doubt anyone has the slightest clue about the global real-world effect of the diverse queing and limiting strategies, especially if you take all the bugs and corner cases into account. Furthermore, everyone has different aims and motivation. Some want to improve fairness whereas there's little agreement about what is fair and what isn't in this context. Others just want to maximize the throughput. -- Christian ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ gtk-gnutella-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/gtk-gnutella-devel
