The Freenet Project

> I want to add something to the HandshakeReply that denotes exactly what build
> is being used, for this type of testing. I tried using Node.class.hashCode()
> but this is different for different executions (as I guess it should be).
> 
> I can't think of anything creative, so that leaves a Build number to bump
> manually everytime one commits. Anyone else have a fresher idea?

Great idea. I can't think of anything myself. It would be nice if it was
automated to rule out forgetfullness, but I can't think of any way to do
it in the code.

Well, things are definitely broken. I'm getting RequestFailed on insert
for a key and also on request for a key, which should never happen.

I don't have time to track it down right now because my conference starts
and I need to pick up my badge. Good luck, debuggers.

Why should that never happen? There is no InsertFailed anymore, remember.
RequestFailed means either that a DataRequest failed to propogate any further,
or that a InsertRequest failed to propogate any further (and again, that means
you can't insert, but you can just lower the htl and start over).

On Fri, 10 Mar 2000, you wrote:
> Well, things are definitely broken. I'm getting RequestFailed on insert
> for a key and also on request for a key, which should never happen.
> 
> I don't have time to track it down right now because my conference starts
> and I need to pick up my badge. Good luck, debuggers.
> 
> 
> 
> 
> _______________________________________________
> Freenet-dev mailing list
> Freenet-dev@...
> http://lists.sourceforge.net/mailman/listinfo/freenet-dev
-- 

Oskar Sandberg

md98-osa@...

#!/bin/perl -sp0777i<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<j]dsj
$/=unpack('H*',$_);$_=`echo 16dio\U$k"SK$/SM$n\EsN0p[lN*1
lK[d2%Sa2/d0$^Ixp"|dc`;s/\W//g;$_=pack('H*',/((..)*)$/)

It should never happen because if I get a RequestFailed on an insert, then
the data is already in Freenet. If I get a RequestFailed on request, then
the data is not in Freenet.

> Why should that never happen? There is no InsertFailed anymore, remember.
> RequestFailed means either that a DataRequest failed to propogate any further,
> or that a InsertRequest failed to propogate any further (and again, that means
> you can't insert, but you can just lower the htl and start over).
> 
> On Fri, 10 Mar 2000, you wrote:
> > Well, things are definitely broken. I'm getting RequestFailed on insert
> > for a key and also on request for a key, which should never happen.
> > 
> > I don't have time to track it down right now because my conference starts
> > and I need to pick up my badge. Good luck, debuggers.
> > 
> > 
> > 
> > 
> > _______________________________________________
> > Freenet-dev mailing list
> > Freenet-dev@...
> > http://lists.sourceforge.net/mailman/listinfo/freenet-dev
> -- 
> 
> Oskar Sandberg
> 
> md98-osa@...
> 
> #!/bin/perl -sp0777i<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<j]dsj
> $/=unpack('H*',$_);$_=`echo 16dio\U$k"SK$/SM$n\EsN0p[lN*1
> lK[d2%Sa2/d0$^Ixp"|dc`;s/\W//g;$_=pack('H*',/((..)*)$/)
> 
> _______________________________________________
> Freenet-dev mailing list
> Freenet-dev@...
> http://lists.sourceforge.net/mailman/listinfo/freenet-dev
>

No, listen man. It used to be that way, when there was InsertFailed message for
backtracking Inserts, so if the Insert couldn't propogate you would get an
InsertFailed. Now (by your suggestion) I use only one backtracking RequestFailed
message, so regardless of whether it is a DataRequest or an InsertRequest, if
the Node can't send the message any further, it will return a RequestFailed.

If the data is in the network you will get a DataReply, not a RequestFailed,
back. Getting a RequestFailed should make the client write: "Your node is
disconnected from Freenet".

And, for the millionth time, if you get a RequestFailed back to the client on a
network that is not just a couple of test nodes, you can be sure that something
is very very wrong.

On Sat, 11 Mar 2000, you wrote:
> It should never happen because if I get a RequestFailed on an insert, then
> the data is already in Freenet. If I get a RequestFailed on request, then
> the data is not in Freenet.
> 
> > Why should that never happen? There is no InsertFailed anymore, remember.
> > RequestFailed means either that a DataRequest failed to propogate any further,
> > or that a InsertRequest failed to propogate any further (and again, that means
> > you can't insert, but you can just lower the htl and start over).
> > 
> > On Fri, 10 Mar 2000, you wrote:
> > > Well, things are definitely broken. I'm getting RequestFailed on insert
> > > for a key and also on request for a key, which should never happen.
> > > 
> > > I don't have time to track it down right now because my conference starts
> > > and I need to pick up my badge. Good luck, debuggers.
> > > 
> > > 
> > > 
> > > 
> > > _______________________________________________
> > > Freenet-dev mailing list
> > > Freenet-dev@...
> > > http://lists.sourceforge.net/mailman/listinfo/freenet-dev
> > -- 
> > 
> > Oskar Sandberg
> > 
> > md98-osa@...
> > 
> > #!/bin/perl -sp0777i<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<j]dsj
> > $/=unpack('H*',$_);$_=`echo 16dio\U$k"SK$/SM$n\EsN0p[lN*1
> > lK[d2%Sa2/d0$^Ixp"|dc`;s/\W//g;$_=pack('H*',/((..)*)$/)
> > 
> > _______________________________________________
> > Freenet-dev mailing list
> > Freenet-dev@...
> > http://lists.sourceforge.net/mailman/listinfo/freenet-dev
> > 
> 
> 
> _______________________________________________
> Freenet-dev mailing list
> Freenet-dev@...
> http://lists.sourceforge.net/mailman/listinfo/freenet-dev
-- 

Oskar Sandberg

md98-osa@...

#!/bin/perl -sp0777i<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<j]dsj
$/=unpack('H*',$_);$_=`echo 16dio\U$k"SK$/SM$n\EsN0p[lN*1
lK[d2%Sa2/d0$^Ixp"|dc`;s/\W//g;$_=pack('H*',/((..)*)$/)

> And, for the millionth time, if you get a RequestFailed back to the client on a
> network that is not just a couple of test nodes, you can be sure that something
> is very very wrong.

That's what I'm saying. I'm getting a RequestFailed from piclab, so I
think something is very very wrong. And this is on both insert and
request, too.

On Sun, 12 Mar 2000, you wrote:
> > And, for the millionth time, if you get a RequestFailed back to the client on a
> > network that is not just a couple of test nodes, you can be sure that something
> > is very very wrong.
> 
> That's what I'm saying. I'm getting a RequestFailed from piclab, so I
> think something is very very wrong. And this is on both insert and
> request, too.

If you get a RequestFailed on a Request, then you will get it on an Insert with
the same HTL. They are routed the same way, so if one runs out of nodes to be
sent to, you can be sure that so does the other.

There is a bug with the handshaking though, looking through the logs Lee
posted, I see things like this:

08-Mar-00 9:48:59 PM:message/Request.java:Minor:Forwarding query to tcp/195.230.51.73:19114
08-Mar-00 9:48:59 PM:Node.java:Minor:Connection attempt timed out
08-Mar-00 9:48:59 PM:Node.java:Minor:Couldn't send handshake request
08-Mar-00 9:56:29 PM:Node.java:Minor:Accepted connection:tcp/216.121.191.70:800

As you can see, it locks up after "Couldn't send handshake request", and just
sits there for 10 minutes. Looking down I can't see it restarting the Query do
to timeout either, so that would lead me assume that it never scheduled the
timeout, ie the thread never finnished. Very weird.



> 
> 
> 
> _______________________________________________
> Freenet-dev mailing list
> Freenet-dev@...
> http://lists.sourceforge.net/mailman/listinfo/freenet-dev
-- 

Oskar Sandberg

md98-osa@...

#!/bin/perl -sp0777i<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<j]dsj
$/=unpack('H*',$_);$_=`echo 16dio\U$k"SK$/SM$n\EsN0p[lN*1
lK[d2%Sa2/d0$^Ixp"|dc`;s/\W//g;$_=pack('H*',/((..)*)$/)