Re: Foundry Networks ServerIron

Well this is what I have found, using the OS that our unit shipped with
from about a month ago. (I believe it was 3.1.05) The following is a
tcpdump, I will try to explain what is going on.
(Kinda long, but it is all useful, atleast I striped the data out :)

TCP Setup

TCP from 207.173.132.110.1099 to 204.71.177.70.80
        seq 586858, ack 0x0, win 8192, SYN

TCP from 204.71.177.70.80 to 207.173.132.110.1099
        seq E483A19E, ack 0x586859, win 17520, SYN ACK

TCP from 207.173.132.110.1099 to
204.71.177.70.80
        seq 586859, ack 0xE483A19F, win 8760,
ACK

Client sending url

TCP from 207.173.132.110.1099 to 204.71.177.70.80
        seq 586859, ack 0xE483A19F, win 8760, PUSH ACK , 404
bytes

Url getting acked plus first data packet

TCP from 204.71.177.70.80 to 207.173.132.110.1099
        seq E483A19F, ack 0x5869ED, win 17520, ACK , 1460
bytes

sec data packet

TCP from 204.71.177.70.80 to 207.173.132.110.1099
        seq E483A753, ack 0x5869ED, win 17520, PUSH ACK , 588
bytes

client acking first packet

TCP from 207.173.132.110.1099 to 204.71.177.70.80
        seq 5869ED, ack 0xE483A19F, win 8760, ACK

server sending the data packets over due to missed ack on 2nd data
packet
(why it is doing the first packet, that was acked, vs the 2nd that was
not? donno)

TCP from 204.71.177.70.80 to 207.173.132.110.1099
        seq E483A19F, ack 0x5869ED, win 17520, ACK , 1460 bytes

TCP from 204.71.177.70.80 to 207.173.132.110.1099
        seq E483A19F, ack 0x5869ED, win 17520, ACK , 1460 bytes

TCP from 204.71.177.70.80 to 207.173.132.110.1099
        seq E483A19F, ack 0x5869ED, win 17520, ACK , 1460 bytes

TCP from 204.71.177.70.80 to 207.173.132.110.1099
        seq E483A19F, ack 0x5869ED, win 17520, ACK , 1460 bytes

TCP from 204.71.177.70.80 to 207.173.132.110.1099
        seq E483A19F, ack 0x5869ED, win 17520, ACK , 1460 bytes

TCP from 204.71.177.70.80 to 207.173.132.110.1099
        seq E483A19F, ack 0x5869ED, win 17520, ACK , 1460 bytes

This happens untill the retry limit is hit.

The browser will timeout and reset the tcp connection, never showing
the
data it was sent.

This only happens on nt RAS connections, NT on the local ethernet did
not have this problem, NT4 with sp3 or sp4 connecting with RAS did have
the problem. Win98 did not have the problem, and I did not have a win95
box to test it with, but since our support people are not screaming at
me, I guess that means it does not have the problem :)

We believe it was the syn flood protection in the foundry causing the
problems. Foundry got us a image later that night, and it seems to be
working now.

I am interested in hearing any other problems people have had with these
boxes.
Received on Mon Nov 23 1998 - 12:10:11 MST