FYI: After running this version for 50 mins, all activity abruptly stopped. Input and output went from 56k/20k to 5k/0k over a period of 2 minutes. No response from localhost:8888 either yet the node was apparently still running as there were dozens of these log entries:
(freenet.node.states.request.ReceivingReply, YThread-1008, NORMAL): Receiving connection died in freenet.node.states.request.ReceivingReply: etc.. followed by quite literally hundreds like this: (freenet.message.DataRequest, YThread-2263, NORMAL): Long messageInitialStateTime 13625 etc... and then a section with the occasional Long messageInitialStateTime interspersed with (freenet.Ticker, Ticker immediate execution thread, NORMAL): On restarting the node, rates have gone back to normal. Lets see how long it stays up... Kevin. -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Toad Sent: 25 November 2003 20:45 To: [EMAIL PROTECTED]; [EMAIL PROTECTED] Subject: [freenet-support] Freenet stable build 5041 Freenet stable build 5041 is now available. Upgrade using update.sh or freenet-webinstall.exe, or get the jar from http://freenetproject.org/snapshots/freenet.jar . Changelog: * Major continuing work on routing: ** Contact the least accessed node every 30 seconds with a recently requested key. This is estimated to add 3kqph to load, but is very important given the next item. ** New nodes in the routing table get pessimistic estimators. So routing is not dominated by new nodes any more. They will get routed to once they have demonstrated their abilities through the above mechanism. ** Try to manage routing table churn: accept new nodes into the routing table from any wierd source if we have less than 40 available, not backed off, nodes. If we have more than that, do probabilistic referencing for StoreData's, and don't borrow refs from requests. ** Impose a network level maximum file size of 1MB. This should make multiplexing easier, simplify code, and prevent some bad behaviours that might be used for traffic analysis. ** Make the global estimate, the request failure time, used in routing, depend on the key AND the filesize. This prevents bad routing of files of non-typical sizes. ** Always reset the datasource when we serve data from the store. ** Other minor changes. * When we do a request with 'skip datastore' enabled, don't actually delete the file from the datastore, just ignore it. * Show the version of each node on the inbound requests page. Improve the open connections and node reference status pages. More diagnostics. Etc. * Lots more bugfixes, logging changes, and other minor improvements. -- Matthew J Toseland - [EMAIL PROTECTED] Freenet Project Official Codemonkey - http://freenetproject.org/ ICTHUS - Nothing is impossible. Our Boss says so. _______________________________________________ Support mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support
