On Fri, Dec 14, 2001 at 02:22:52PM -0500, Gianni Johansson wrote: > On Friday 14 December 2001 09:35, Toad wrote: > > Hi. According to localhost:8889, my (severely borken) node has 9 > > references. The top 6 have CP = 1.0 and have never been contacted. > Never successfully contacted or never tried? > > There's a time based recovery scheme that causes the CP of nodes that haven't > been retried in a while to drift up to 1.0. > > > I am > > running with debug logging, so I grepped my logs. Four of them don't appear > > at all, ever, anywhere in the log (for today). > Don't know. Maybe they were in the list of seedNodes that your node was > seeded with. > > > Node has been running for > > about 8 hours, with heavy insertions (mostly failed). > The network is fragile if you do too many insertions too fast you will > overwhelm all your noderefs. Ok, it's not that heavy. Light insertions, one file at a time using fcpput. Small files. Maybe 35 of them, plus another load that should collide with stuff previously inserted. > > >Surely this is a bug in Fred? > A bug implies only only one. I appreciate your optimism! :) > > I have also seen cases where nodes where never even though not for several > weeks. It looked like it might be the result of a corrupted data store. Yeah, mine has gone from 67 keys in the datastore and 8 contacted noderefs, none of which worked reliably, to 1 key and 1 noderef, last attempt 20 minutes ago... in the space of another few hours :( Datastore bug fix will be a wonderful xmas present, but maybe not the only thing at work here... > > Also, there were bugs in the very first versions of NodeStatusServlet (like a > month ago) that might have given you this behavior. So if you are running an > ancient build, all bets are off. As I pointed out in another mail, I use daily updated freenet.jar from the snapshots dir.
_______________________________________________ Devl mailing list Devl at freenetproject.org http://lists.freenetproject.org/mailman/listinfo/devl
