Re: Linux 2.2.14 TCP problem ??

From: David S. Miller (davem@redhat.com)
Date: Wed Jan 05 2000 - 16:14:38 EST


   From: Rob van Nieuwkerk <robn@verdi.et.tudelft.nl>
   Date: Wed, 5 Jan 2000 21:41:05 +0100 (CET)

   It's very easy (on my machine) to reproduce: just visit the FTP
   site with netscape and do some cd or read a file.

Easy to reproduce here as well, I believe it is the applix machine's
fault, watch:

 < applix100.ftp > pizda.32908: P 2633:2685(52) ack 1224 win 8760

Applix says "Here is data up to and including 2685"

> pizda.32908 > applix100.ftp: . 1224:1224(0) ack 2685 win 32696 (DF)

Linux says "Ok, I got up to and including 2685"

> pizda.32908 > applix100.ftp: P 1224:1280(56) ack 2685 win 32696 (DF)

Linux says "Here is data up to and including 1280"

 < applix100.ftp > pizda.32908: . 2679:2679(0) ack 1224 win 8760 (DF)

Applix says "I didn't get that"

BUT also note how Applix has decreased it's sequence numbers to before
the data Applix sent before, this is completely illegal, you cannot
take back what you've already sent to the other end and the other end
has acknowledged.

I note in passing that this machine is a Solaris system. But it is
also possible that some load balancing "black box" is at work here,
I've seen in the past cases where such devices mess with TCP headers
and cause no end of headaches for properly functioning TCP
implementations (the worst on record is hotmail.com).

Later,
David S. Miller
davem@redhat.com

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Fri Jan 07 2000 - 21:00:04 EST