[1446] in linux-net channel archive
Re: TCP getting confused
daemon@ATHENA.MIT.EDU (Greg Stein)
Wed Nov 29 06:20:25 1995
Date: Tue, 28 Nov 1995 14:06:17 -0800
To: Alan Cox <alan@cymru.net>
From: greg_stein@eshop.com (Greg Stein)
Cc: linux-net@vger.rutgers.edu, linux-kernel@vger.rutgers.edu,
greg_stein@eshop.com
Hey Alan... thanx for responding!
At 5:41 AM 11/28/95, Alan Cox wrote:
>> Here is where things freaked out. The next packet Linux sent was at 28.440
>> with seq #45934. Note that a packet is missing. This was immediately
>
>Ok that in itself is fine, one got lost on cable or on the router.
My sniffer was on the same segment as the linux machine. It should have
seen something, I thought... (it does record runt packets and frame errors
and stuff, but can't really see collisions).
>> At 28.443, the Sun resent four ack packets for #45486. At 28.858, the Linux
>> machine again sent packet #35630. From this point on, the linux and sun
>> traded acks for 45486 and sending pkt 35630. This went on for a couple
>> minutes. I stopped the FTP at this point.
>
>Do you have a BOCA PCI ethernet card in one of those boxes ?
Nope... As I mentioned in my original note, the Linux machine was using an
Intel EtherExpress Pro (PnP disabled, using int10, 0x300). The router also
uses EtherExpress Pros. All of these are ISA cards.
>> another host, but nothing went out. Checking the ARP cache, I found this
>> new host had an address of 00:00:00:00:00:00. Ick. Lastly, I also watched
>
>Thats fine. Its trying to resolve an entry. It just hasnt completed yet.
It never completes. It seems as if the data receipt just doesn't work
(based on my ping tests as described below).
>What hardware is first question - including the cards on the MPR router
I'm not sure what level is relevant, so I'll try to guess:
Pentium 133, Intel motherboard (not sure of the type). 32 Meg ram.
Two WDC IDE drives. One 4x speed ATAPI CDROM (Mitsumi).
SB16 clone card. Calcomp 14.4 internal on cua2.
ATI Mach64 video card. AMI BIOS.
Nothing of interest was running on the hardware at the times I reproduce
this; atalkd and afpd might be the only unusual ones. In my original note,
I guessed they may have gone offline when I "downed" the interface; I found
they went down long before that. Based on the DDP also going down, it seems
to point to something other than the TCP protocol stacks.
Some more information since yesterday:
- I had a valid ARP entry for another machine on the same segment, so I
pinged it. My sniffer saw the ping go out and a response come back. My
machine did not see it.
- I tried two EtherExpress Pro cards, so I doubt it is the card. Even
weirder, the system seemed to be working fine up until yesterday (for
several weeks).
- I gave up after a while, yanked the Pro and plugged in a 3C503, and
rebuilt a 1.3.45 kernel for it. Works great now.
If there are some specific tests and/or data that you need, I still have my
old kernel and I have no problems swapping cards if needed to track this
down. Let me know how I can help.
Thanx much,
Greg
p.s. I'm still copying the mailing lists, as maybe somebody has an idea or
has seen this, but I might guess that this problem might not continue to
need the distribution(?)...
Greg Stein, eShop Inc.
greg_stein@eshop.com
