[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[linrad] Re: Testing MAP65 v0.8



Rick, and all --

OK, I've a chance to make some more tests.

My computer network looks like this:


        ADSL      10 Mb/s  --> Computer_A
DSL --> Modem --> Ethernet --> Computer_B
                  Hub            |
                           --> Computer_C

Three computers are connected to a 10 Mb/s Ethernet Hub.

Additional information: "ipconfig" on Windows, "ifconfig" on Linux, report the following IP addresses:

Computer A:    172.16.28.67
Computer B(1): 172.16.28.69
Computer B(2): 192.168.10.13
Computer C(1): 172.16.28.31
Computer C(2): 192.168.10.12

Have you considered replacing the hub with a 100 Mbps full-duplex Ethernet switch? There are many advantages in this over a hub.

Yes. That was my first attempt at a solution. I tried replacing the 10 Mb/s hub with a 10/100 Mb/s switch. The result was the same: when Computer C was multicasting 16-bit Linrad data at about 0.77 MB/s, Computer A was essentially unable to use the internet. The switch apparently did not prevent multicast traffic from reaching A.

This was with a "D-Link 10/100 Desktop Ethernet Switch. I also tried it with a Linksys model EZXS55W "EtherFast 10/100 5-port Workgroup Switch." Same result.

I then tried using both the hub and the switch:

         ADSL      10 Mb/s  --> Computer A
 DSL --> Modem --> Ethernet
                   Hub      --> Ethernet --> Computer_B
                                Switch          |
                                         --> Computer_C


Again, no change. This time I checked and confirmed that packets were arriving at A at the correct rate for them to be the multicast packets from C.

Computer_A is my XYL's machine. Computer_B runs Windows 2000 Pro, and Computer_C runs Linux (presently the Kubuntu 6.06 distribution). In addition to the connections of all three machines to the hub, a crossover cable makes a direct 100 Mb/s connection between computers B and C.

The ethernet interfaces on B and C appear to be configured correctly. On Linux they appear as eth0 and eth1 (occasionally they boot up as eth0 and eth2, I don't know why???).

This is configurable, generally, and should be fixed if you intend to use interface based static routes. Check here for more info on iftab (/etc/iftab):

http://linux.die.net/man/5/iftab

RRR, thanks.

 Connections to the Hub are assigned dynamic IP addresses;

I assume these addresses are in the 192.168.1.x range?

No, see above. I was probably wrong to call them dynamic IP addresses. They are assigned by DHCP, but I believe they are always the same.

I assigned hard-coded addresses 192.168.10.12 and 192.168.10.13 for the direct inter-machine connection
between B and C.

I can use the 100 Mb/s direct line for many purposes. I can ping over it in either direction; I can ssh into Linux from Windows; I can use Cygwin/X (as described above) to display Linux X programs on the Windows screen.

However, I cannot seem to persuade Windows 2000 Pro to accept multicast packets over the direct line. When I run Linrad on computer C and MAP65 on B, the multicast traffic is always received over the slow line, through the Hub. This uses most of the 10 Mb/s link's bandwidth, and my wife can't read her email when I'm on the air. This is NOT GOOD.

An Ethernet switch would eliminate this, as traffic passing between two machines (B-C) does not use any bandwidth, nor is it seen, by any other machines. Internet access by machine A would be unaffected by a transfer occurring between machines B and C. Machine A would not see the traffic, nor would there be any contention for bandwidth on it's connection because of the B-C traffic.

Well, as far as I can see this does not seem to be the case. Can it be that your statement is true for normal one-to-one IP traffic, but not for multicast traffic? Or is it true for a router, but not for a switch?

By default the multicast traffic generated by Computer_C goes to eth0. I can use the Linux "route" command to explicitly tell the system to use eth0:

# route add -net 224.0.0.0 netmask 224.0.0.0 dev eth0

This works fine (but of course, still sends the heavy multicast traffic through the hub). If I remove this routing instruction and instead enter

# route add -net 224.0.0.0 netmask 224.0.0.0 dev eth1

the multicast data are not received by MAP65 running on the other machine.

If I unplug the crossover cable from the Windows machine and instead plug it into a laptop running Win/XP, the laptop receives the multicast packets without a problem.

Thus, it would seem that the problem must be in my setup of the Win2k machine -- the one with two ethernet interfaces. Can anyone shed any light on this situation for me?

Would there be sufficient bandwidth in a 100baseTx connection (100 Mbps full-duplex) to handle both of the networking streams, i.e. the hub and the direct stream? If so, replacing the inefficient hub with a faster switch, thus confining network traffic to only the ports of the involved machines, might solve the issue. This might allow you to eliminate the direct connection between machines B and C.

Yes, this would be fine ... if it worked. So far, I have not been able to make it work.

Anyway, the direct line between B and C should work, no? Why can't I persuade B to receive multicast packets from C on the direct line??

	-- 73, Joe, K1JT

#############################################################
This message is sent to you because you are subscribed to
 the mailing list <linrad@xxxxxxxxxxxxxxxxxxxxx>.
To unsubscribe, E-mail to: <linrad-off@xxxxxxxxxxxxxxxxxxxxx>
To switch to the DIGEST mode, E-mail to <linrad-digest@xxxxxxxxxxxxxxxxxxxxx>
To switch to the INDEX mode, E-mail to <linrad-index@xxxxxxxxxxxxxxxxxxxxx>
Send administrative queries to  <linrad-request@xxxxxxxxxxxxxxxxxxxxx>

LINRADDARNIL