lkml.org 
[lkml]   [1996]   [Mar]   [31]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Slow PPP - caused by duplicate packets?
At 08:10 PM 3/30/96 +0200, Linus Torvalds wrote:
>
>Umm.. The tcpdump info you include would seem to indicate that linux is
>doing everything correctly, and that the Solaris box is doing something
>stupid.
>
>However, I can't imagine that Solaris started doing stupid things for
>everybody in the 1.3.7x timeframe and worked fine before that, so there
>must be something we're not seeing here. Does anybody have access to
>tcpdump at the _other_ end?
>
>It _looks_ like the other end is ignoring our first ACK, so it times out,
>and then sends the packet again. And then we ACK it twice (that's normal
>linux behaviour - there's the "normal" ACK, and then there is an extra
>one that comes from the fact that the linux TCP code decides we'd better
>ack again when we got a packet re-sent).
>
>Now, WHY is Solaris ignoring our first ACK?
>

Some more data...I've partially figured this out...

Here's my setup:

Linux box1 -> ethernet -> Linux box2 -> PPP -> Sun OS (UNIX(r) System V
Release 4.0)

I had the dup packets occurring between box2 and Sun for the last several
weeks, but only on FTP get (i.e. Sun to Linux) and not on puts.

This morning I restarted PPP with "mru 576 mtu 576" and the problem went away.

Now, box1 has the same problem ftp'ing to the Sun...puts work fine but gets
have dup packets! FTP's to anything beyond the Sun work fine. I'd not
tried testing from a second box before.

It seems like once the FTP session starts the double ack's it keeps on
doubling. Once in a while it will do a completely clean transfer (but not
very often).

Here's a portion of the tcpdump from the Linux box2 looking at the PPP
connection where the problem starts (ddi is the Sun, medusa is the Linux
box1). As you can see at 39.938957 ddi sends 1073 (2nd packet) for a second
time (as though the first ack for this packet got lost). Then, at 39.738957
medusa sends an ack for 2681, then sends another at 39.938957 (exactly
200ms!?clue). DDI responds by resending 2681 again at 40.648957 (3 packets
later). Then we get double and triple acks for the rest of the FTP. I've
also tried ftp'ing to sunsite (which is also a Sun 4.0 box) and that works
OK...so this problem only seems to apply to the PPP host. Also, at
39.938957 when 1073 is received for the 2nd time, shouldn't medusa ACK 1073
instead of 2681?

07:40:38.838957 ddi.digital.net.20 > medusa.csihq.com.1118: P 1:537(536) ack
1 win 25192 (DF)
07:40:38.838957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 537 win 12864
07:40:38.858957 medusa.csihq.com.1114 > ddi.digital.net.ftp: . ack 371 win 12955
07:40:39.208957 ddi.digital.net.20 > medusa.csihq.com.1118: . 537:1073(536)
ack1 win 25192 (DF)
07:40:39.208957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 1073 win 12864
07:40:39.368957 ddi.digital.net.20 > medusa.csihq.com.1118: P 1073:1609(536)
ack 1 win 25192 (DF)
07:40:39.368957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 1609 win 12864
07:40:39.538957 ddi.digital.net.20 > medusa.csihq.com.1118: . 1609:2145(536)
ack 1 win 25192 (DF)
07:40:39.538957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 2145 win 12864
07:40:39.738957 ddi.digital.net.20 > medusa.csihq.com.1118: P 2145:2681(536)
ack 1 win 25192 (DF)
07:40:39.738957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 2681 win 12864
07:40:39.938957 ddi.digital.net.20 > medusa.csihq.com.1118: . 1073:1609(536)
ack 1 win 25192 (DF)
07:40:39.938957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 2681 win 12864
07:40:40.108957 ddi.digital.net.20 > medusa.csihq.com.1118: P 2681:3217(536)
ack 1 win 25192 (DF)
07:40:40.108957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 3217 win 12864
07:40:40.278957 ddi.digital.net.20 > medusa.csihq.com.1118: . 3217:3753(536)
ack 1 win 25192 (DF)
07:40:40.278957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 3753 win 12864
07:40:40.448957 ddi.digital.net.20 > medusa.csihq.com.1118: P 3753:4289(536)
ack 1 win 25192 (DF)
07:40:40.448957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 4289 win 12864
07:40:40.648957 ddi.digital.net.20 > medusa.csihq.com.1118: . 2681:3217(536)
ack 1 win 25192 (DF)
07:40:40.648957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 4289 win 12864
07:40:40.778957 ddi.digital.net.20 > medusa.csihq.com.1118: . 3753:4289(536)
ack 1 win 25192 (DF)
07:40:40.778957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 4288 win 12864
07:40:40.778957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 4288 win 12864
07:40:40.978957 ddi.digital.net.20 > medusa.csihq.com.1118: . 4288:4824(536)
ack 1 win 25192 (DF)
07:40:40.978957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 4824 win 12864
07:40:41.148957 ddi.digital.net.20 > medusa.csihq.com.1118: . 4824:5360(536)
ack 1 win 25192 (DF)
07:40:41.148957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 5360 win 12864
07:40:41.338957 ddi.digital.net.20 > medusa.csihq.com.1118: P 5360:5896(536)
ack 1 win 25192 (DF)
07:40:41.338957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 5896 win 12864
07:40:41.518957 ddi.digital.net.20 > medusa.csihq.com.1118: . 4288:4824(536)
ack 1 win 25192 (DF)
07:40:41.518957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 5896 win 12864
07:40:41.638957 ddi.digital.net.20 > medusa.csihq.com.1118: . 4824:5360(536)
ack 1 win 25192 (DF)
07:40:41.638957 medusa.csihq.com.1118 > ddi.digital.net.20: . ack 5896 win 12864
07:40:41.768957 ddi.digital.net.20 > medusa.csihq.com.1118: . 5360:5896(536)
ack 1 win 25192 (DF)
-
/----------------------------------------------------------\
| Mike Black mblack@csihq.com 407-676-5118, x203 |
| Computer Science Innovations, Inc. 407-676-2355 FAX |
| Melbourne FL 32904-2314 http://www.csihq.com |
\----------------------------------------------------------/



\
 
 \ /
  Last update: 2005-03-22 13:36    [W:0.032 / U:0.400 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site