Florent Daigni?re wrote:

>> This also provides a possibility for a denial of service attack against
>> the 0.7 network - from the code it is not too difficult for someone
>> wishing to impede free speech (or the 0.7 against 0.5 networks) to
>> create bogus nodes that report way too high ping times. This would
>> severely hurt the 0.7 network from reaching its potential.
> 
> Using the median insteed of the mean solves that problem... Moreover
> really high ping times are not possible because there is a timeout.

Actually, the problem is worsened by using the median. After all, the
median becomes much faster high than the average if there is only 1 peer
that has a high ping time and 10 that have a low ping time.

The main problem with my node appears to have been the use of both ipv4
and ipv6 over the tunneling link on a NAT -system. which caused for some
reason the node to receive packets twice and eding up with huge ping times.

However, I have occasionally see situations where the averagepingtime
increases because of one peer having problems. Of course, the node can
be cut off, but that sort of manual maintenance should be kept to a
minimum if possible.

Using minpingtime (at least as an alternative, might cause problems if
two nodes are behind the same DSL etc.) would solve the problem, and as
pointed out, would also more accurately describe the load of a single node.

-- 
Malkus Lindroos



Reply via email to