[Dxspider-support] "you are connected to me" Problem

Lee Sawkins ve7cc at shaw.ca
Thu Dec 30 11:07:30 GMT 2004


Hi Dirk.

Thanks for explaining what was happening.  It makes a bit more sense
now.

I hate to sound like I am pushing my program all the time, but this is
how my program works.  When it gets an unwanted disconnect, it waits one
minute and then reconnects.  If it gets disconnected a couple of times
in a few minutes, it then generates a random SSID and reconnects.  If it
gets disconnected again, it reverts back to the original SSID and
reconnects and the cycle will repeat forever.  If someone wants to keep
you disconnected, they have to really work at it.  Since my program
pulls in all the missed spots, the user isn't really aware this is even
happening unless he happens to be watching the link.

This feature of my program was not put in there for to prevent your
enemies from causing you a problem.  It was put there for users of my
program that don't realize you can't make 3 connections to a node with
their 3 computers, all of which are using their callsign without an
SSID.  It also works with those who connect at home and then leave it
connected full time.  When they go to work they also connect, using the
work computer.  This features applies an SSID to one of the links, and
all is happy!  It also works where not all the computers are running my
program.  The ones that are running my program will change SSID and
leave the other one with its original SSID.

One thing that Spider does that I don't like is that it treats all these
SSIDs are completely separate users.  They have to log in and resend all
their information.   The big problem is that the filters don't carry
over from one SSID to the other.  In AR Cluster the filters are carried
from one SSID to another, unless the SSID is over 15.  That is why you
occaassionally see AR Cluster users with SSIDs of -16, -17 etc.

Lee VE7CC

Dirk Koopman wrote:
> 
> On Tue, 2004-12-28 at 19:20 -0800, Bob Nielsen wrote:
> > On Tue, Dec 28, 2004 at 04:51:20PM -0800, Lee Sawkins wrote:
> > >
> > > My user program detects the "you are connected to me on another port"
> > > message.  It then generates a random SSID and reconnects.  G3VQO is
> > > using this program.  Today G3VQO and G3VQO with 11 different SSIDs were
> > > connected to me at the same time.  That's correct, 12 connections.  The
> > > original link took 8 hours to disconnect after it quit working.  The
> > > others took various amounts of time.  As I have said before, why not
> > > simply disconnect the original link when a new one shows up with the
> > > same call?
> >
> > One reason I can see is that someone who wanted to cause trouble could
> > disconnect other users that way.
> >
> > With links there is a mechanism using ping to cause a disconnect when
> > the obscount is exceeded.  Perhaps a ping could be used for users as
> > well as links to other clusters.  This wouldn't be immediate, but would
> > be much faster than 8 hours.  An obscount setting would allow a bit of
> > network flakiness.
> 
> This is a problem with the config of some ISPs that force dynamic
> addressing onto their users. It exposes a "feature" in the IP protocol
> itself. Because it is a datagram protocol, packets are just sent out, no
> reply received need be received and this is regarded as normal.
> 
> When someone on a dynamic address gets a new one, because they have been
> disconnected after 2 hours or whatever, that address goes back into the
> pool. If it is not reassigned (to a new punter), then the IP stack on
> the node will not notice that it is gone until the maximum window size
> (usually 32K) has filled up. On a slow day this can take ages.
> 
> However, on a busy ISP, usually these addresses are recycled fairly
> quickly. When the new owner gets a packet on a circuit that he is not
> expecting he will send an ICMP message saying, in effect, "I don't know
> you" - go away. The IP, stack on the node, picks this up as a
> "disconnection", that channel in the node is then cleared down and he
> can then reconnect with his callsign.
> 
> Now, you may say "why not disconnect the old call if he tries again",
> the answer to that is "to prevent nastiness". Basically, one of the
> joyous things one could then do is: see your mortal enemy on a node and
> disconnect him, simply by logging in on the same node (and then
> disconnecting of course).
> 
> Now this was a problem on ax25 nodes and it has happened in the past on
> telnet based nodes - hence the restriction.
> 
> However, on further thought, if you have passwords then this restriction
> could be removed. Another alternative would be a configuration option
> that allows say telnet users to bump themselves off and automatically
> reconnect.
> 
> Please discuss.
> 
> Dirk G1TLH
> 
> _______________________________________________
> Dxspider-support mailing list
> Dxspider-support at dxcluster.org
> http://www.tobit.co.uk/mailman/listinfo/dxspider-support



More information about the Dxspider-support mailing list