In message <[EMAIL PROTECTED]>, Oskar Sandberg <[EMAIL PROTECTED]> writes
On Sun, Nov 17, 2002 at 09:38:45PM +0000, Roger Hayter wrote:
<>
I find that surprising.  I am not sure what you mean by "sending about
300 pieces of data per hour".  My node deals with about 150 to 300
queries per hour in terms of "incoming queries that are not rejected".
Or do you only include queries which were successful?  Or only ones
where the data was on your node (I don't know how to find this)? "Number
of times data was sent" is less impressive at 6 to 30 times per hour -
is this the parameter you were referring to? But my node runs on a 70MHz
pentium, uses 12-16MB of RAM, and has an outbound bandwidth of 2.5kb/s
(I assume the bandwidth limits are kb/s not kByte/s!).  This is build
534, which seems remarkably more efficient than most, apart from
occasionally losing all its node references.
I was referring to the sentData field, yes. That shows the number of
times that data was sent from the node (either when transfering or an
insert or data request, or when data was found on it). The number hovers
between 300 and 500 per hour here. That is for about 8000 accepted
queries - about 96% of the queries my node sees fail (I have no idea
where they all come from, but I understand there is some software that
floods the network).

As for you 300 qph is nothing. I would start wondering about the load
balancing, but I'm surprised that freenet even runs on 70 mhz, so...

Seems to cope alright with the bandwidth I allow it! It is surprising that all the feedback controls in Freenet nodes reach any kind of stability over a range of different node and network states, but my present combination seems to work fine, and make a small but valid contribution, except for occasionally losing all its node references.
--
Roger Hayter

_______________________________________________
devl mailing list
[EMAIL PROTECTED]
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to