toad <matthew at toseland.f9.co.uk> writes:

> On Fri, Dec 14, 2001 at 09:13:41AM -0600, Edgar Friendly wrote:
<SNIP>
> > The reason you have nodes with high CP that haven't been contacted is
> > that fred already punished them (by not talking to them for a while),
> > and is trying to make sure they're still down.  If we left their CPs
> > low, it'd practically never try to talk to them again.
> Hmmm. How is this recorded?

You'll have to ask GJ that, he's the one that coded it.  I think there's some 
sort of 'temporarily at CP=1' flag that's set.
> > 
> > As for why your node is so crappy: it hasn't found even a couple good
> > nodes to talk to.  The best way I detect this is by requesting
> > non-existent keys at varying HTL and seeing how long it takes/how high
> > of a HTL I can get a DNF instead of a RNF.
> It has one, but not so good. RNFs dominate DNFs.

You don't get new noderefs by inserting.  The way you get new noderefs
is successful requests.  You should try requesting a bunch of known
stuff (use frost for a little while), and that'll get you new good
noderefs.

> It has 37 files in a 1.8GB datastore, having run for weeks with daily inserts
> of smallish files, and lots of requests. So I suspect datastore damage.
> Is it possible to reset the datastore and keep the same public key/port?

Yes, --resetDS will do this, but hobx is partially correct in saying
that you should make a new key/port because you're not really the same
node for routing purposes, but I think that at this point, it's
important to just keep references to other nodes even if the routing
info is wrong.  The only kind of datastore damage I'm aware of is the
bug that keeps your node from starting up because the routing table
has been overwritten because your store got full.

> > 
<SNIP>
> > surmise you're still running old node code.  Try downloading a newer
> > snapshot (one from the last couple days should do) and using it.
> It is updated daily from 
> http://www.freenetproject.org/snapshots/freenet-latest.jar ... is this up to 
> date?

I have no idea.  I just use the CVS tree.  There's been some evil
problems in the snapshot-building process in the past, maybe they're
popping up again.

Thelema

_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://lists.freenetproject.org/mailman/listinfo/devl

Reply via email to