Ok...despite Ian telling me to f- read what toad
wrote, what I feared would happen has happend: whether or not I let it run
longer, it's still the same. Basically, it's still crap.
On the contrary, it seems to get worse the longer I
let it on. In the very beginnings, it worked (more or less), but then it went
steadily worse, untill nothing was retrievable. After some discussions (and the
well-thought of responses on the maillist ;-) ) someone on IIP suggested to
update my seednodes.ref. So I did. And indeed, when restarting things seemed to
have improved.
I thought the seednodes didn't update when using
the auto-update, but toad assured me they would/should. And I guess he might be
right, because after a while it deteriorated again, untill it was as crappy as
before.
And that is really crappy: my statusbar is around
1-2% and stays there, no activelinks show up at all, etc.
Now, I AM using a router and what not, but it never
deteriorated like this before, and build 5085 is supposed to work *better* with
routers, right? So, I have basically no idea what is going on (neither do the
coders, I suspect ;-)), but I'm pretty sure it's not due to being not
well-connected to the network in the normal sense and that it will improve with
time.
I would send some data on it, if I knew what
exactly would be interesting to know. (openconnections?)
Connections open (Inbound/Outbound/Limit) |
6 (0/6/200) |
Transfers active (Transmitting/Receiving) |
1 (0/1) |
Data waiting to be transmitted/received |
None/None |
Amount of data transmitted/received over currently open connections |
2,219 KiB/3,740 KiB |
Total amount of data transmitted/received |
24 MiB/27 MiB |
Number of distinct nodes connected |
6 |
After a whole night, 24MB seems rather
puny.
----- Original Message -----
Sent: Friday, July 23, 2004 11:31
AM
Subject: new stable
Tried it out, and thusfar...it's
crap.
Maybe (hope so) this is temporary, but it
definately works WAY worse then it used to. Actually, I almost don't get
anything exept RNF, DNF and other 'can't connect' messages.
I know it's maybe too early to tell, but what if
all the changes didn't do a thing (again) to improve Freenet? I mean...we
could be busy untill men land on mars and Bush becomes a saint (I would've
said an intelligent dude, but I want to keep it remotely
possible).
If things don't work, and don't keep working,
maybe we should make a dramatic departure and throw away the concept of NGR
and the current bw/limiting shemes?
Maybe we just made it too complicated, and we
should revert or at least think about moving it to something more simple
again, akin to the classical routing?
Also, IMHO, one might implement something that
makes a node more keen on (more) rapidly using other nodes it
gets aware of, even if they seem - or are - 'less good' then the
seednodes. I think the seednodes.ref causes rather an all-to-long-remaining
bottleneck when contacting the network.
Maybe it will get (preferable much) better, and I
truelly hope so, but if it doesn't it has been yet another giant waste of
time. I mean, I'm sure there are lots of improvements, but if endusers don't
see the difference and still get a crappy working network, it's rather
pointless.
Maybe I'm just talking in a mood of
dissapointment, but, even as a fan of Freenet (or I wouldn't do what I'm
doing), I'm getting rather tired of all those months and even years of
'improvements' that just don't seem to materialise into a better, faster
Freenet. I've expressed my opinion before: I think the only chance of actually
getting somewhere, is by creating a large, 'real-life' testnetwork where the
actual workings can be followed in detail. Toads' little network already
showed some promise, but falls far short of what is needed.
I can't but feel that, if we had done that a year
before, we would now be much further in the development and with a good chance
of having a working Freenet.
|