On 25/07/15 09:47, Florent Daigniere wrote: > On Fri, 2015-07-24 at 20:19 -0400, Steve Dougherty wrote: >> 1468's release coincides with increased overload probe response - see >> attached. Any ideas on why this might be? Failed upgrades maybe? >> Network >> size estimates are dropping as well but that's a longer-term trend. > My guess is that it's removed a bottleneck somewhere and now the next > one becomes visible. Here my node uses more bandwidth than before and > does process more requests than it used to; to me that explains why the > load-limiting system kicks in and the node returns OVERLOAD more often. > > I've noticed that the top local preemptive rejection reason is "thread > based"... which I think is *very* wrong. Is that a local artefact or > something seen on other nodes too? So far I've found 3 persons on IRC > whose node exhibit the same problem. Is it specific to high bandwidth nodes? I'm not seeing it.
Also I have no idea why we are seeing better network level performance, as appears to be widely believed. Could be because of a large loss of nodes but it doesn't look like it; there is some loss, and it seems to have got worse since June, but it's not that rapid. Could be because of a large loss of *active requestors*, but again we should know about that, there are only sporadic reports and it should be fixed now...
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Devl mailing list [email protected] https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
