Igor:
        Yes, we should skip using subnet masks and return
again to hard coded 254 IPs. Except, if Kaboodle sniffs an
"alien host" (ie, a device with an IP address out of the
Kaboodle PC's subnet range), it should ping that alien's
class-C range as well.

        There's two good reasons to do it this way. First, many
subnet masks set by hand are incorrect, and I wouldn't want
Kaboodle to break for a silly reason. Windows file sharing on
Win98 breaks, for instance, if the machines are not on the
same subnet. Secondly, if a malicious user was sneaking onto
a wireless-LAN, they'd most likely *try* to use a subnet mask
that made it difficult for well-behaved applications to detect
it.

        After this change, network detection should be all
done. Please confirm.

        After this, I want to move forward with the "Master
Node" concept: only one PC on the LAN should be the pinger,
all of the others should be passive sniffers. The Master
Node should also be the one machine that all Zebedee traffic
goes thru during a VPN connection.

thanks,
Scott


On Tue, 17 Sep 2002, Igor Kotelevsky wrote:

> Hello, Scott.
> > Heya. Your most recent changes into the CVS server
> > do not seem to be working. At startup, Kaboodle sends just
> > one ping, to its default gateway. That's it.
>
> Sorry, was a bug. Fixed now.
>
> > > >Okay, lets try it this way. Every Kaboodle instance knows the
> > > >IP address of the other machines on the LAN. Kaboodle should ping
> > > >the *fewest* number of class-C address ranges to get them all. At
> > > >startup, it pings it's own class-C range.
> > >
> > > You obviously mean class-C or smaller (MaskLen >= 24)
> >
> > No, I don't mean smaller. I mean that Kaboodle should
> > ping a.b.c.0 to a.b.c.254 if its IP address is a.b.c.d. Kaboodle
> > only sends this "broadcast" ping at startup and when the user
> > hits Refresh, so I don't mind it being bigger than needed.
>
> But this will mean we refuse using subnet masks and return again to
> hard coded 254 IPs.
>
> > In general, at startup, Kaboodle should:
> >
> > 1. Draw the GUI, put up "Detecting Network..." messagebox.
> > 2. Determine its active interfaces.
> > 3. Iterate thru its active interfaces, and ask other Kaboodle
> >    instances on LAN for a NID.
>
> So far is implemented already
>
> > 4. If #3 fails, it should iterate thru its active interfaces
> >    and ping the 254 addresses of its own interfaces's Class-C
> >    range.
>
> Ping 254 IPs in each network (interface)
>
> > 5. After #3, run the OS-detection thread.
> >
> > When the user hits the "Refresh" button, Kaboodle
> > should:
> >
> > 1. Determine its active interfaces.
> > 2. Iterate thru the active interfaces and ping the 254 addresses
> >    of its interface's own Class-C range.
>
> The same as at startup
>
> > 2. Ping any other Class-C ranges needed to encompass all of
> >    the known devices in the NID. So if Kaboodle knows about
> >    10 machines from 10.0.1.1 thru 10.0.1.10 and one machine
> >    at 10.0.2.1 and another at 10.0.3.1, Kaboodle should send
> >    out 762 pings when the user hits refresh.
>
> In order to clear alien hosts (those my PC is not in one network).
>
> >
> > > >I'm not sure if this will work. Some people I know use
> > > >USB-to-ethernet converters to connect to a LAN. I don't know, but
> > > >I suspect that these interfaces will show up as "MIB_IF_TYPE_OTHER"
> > > >and be ignored. You may want to install just the driver for one
> > > >of these devices (http://www.dlink.com/products/wireless/dwl120/)
> > > >and see what your code thinks of that interface.
> > >
> > > I have tried but it would not work w/o the device. So no way of knowing
> > > how the interface type would be detected.
> >
> > So, what did not work? Did the driver not install, or
> > did Kaboodle stop working?
> >
>
> Kaboodle went on working. The driver installed but would not operate w/o
> hardware. And so no IP was assigned.
> In fact it does not matter as I have implemented you algorithm where
> interface type is not considered.
>
> -Igor
>
>




-------------------------------------------------------
This SF.NET email is sponsored by: AMD - Your access to the experts
on Hammer Technology! Open Source & Linux Developers, register now
for the AMD Developer Symposium. Code: EX8664
http://www.developwithamd.com/developerlab
_______________________________________________
Kaboodle-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/kaboodle-devel

Reply via email to