Re: ICMP in 2.2.9 (was 3c575)

Dennis (dennis@etinc.com)
Sat, 12 Jun 1999 15:48:59 -0400


At 05:00 AM 6/11/99 +0200, jakob@ostenfeld.dk wrote:
>On Fri, Jun 11, 1999 at 03:26:56AM +0100, Alan Cox wrote:
>> > Can anyone confirm, that it is not possible to ping machines running
>> > the 2.2.9 kernel with ping packets of size 1475 ?
>>
>> Yes. Just fine. You probably have a buggy ping binary.
>
>But it's a buggy ping binary on three different releases of redhat
>on four different kernels then.
>
>I do ping -s 1475 some.host.somewhere and get no replies.
>
>This is getting weirder by the minute... Perhaps I should get some
>sleep soon...
>
>
>Host 1 (2.2.7, RH5.9, 3com905)
>Host 2 (2.2.3, RH5.2, Tulip)
>Host 3 (2.0.36, RH5.1, Tulip)
>Host 4 (2.2.9, RH6.0, Tulip)
>
> 1 2 3 4 (destination)
> ---------------------
>1| N N N
>2| Y Y Y
>3| Y Y Y
>4| N N N
>(source)
>
>Yikes. The pattern as I see it is, ping with glibc2.1 cannot
>send _good_ ICMP echo requests.
>
>The really funny thing is, that where you see a `N', eg. when host
>1 pings host 3, the echo request packets actually reaches host 3,
>but host 3 never sends a reply. (Seen with tcpdump on host 3)
>I guess the kernel discards the requests for some reason (??)
>
>
>Unfortunately I only have RedHat boxes around here, so I can't test
>whether it's just RedHat's ping in 5.9 and 6.0 that's screwed, or
>it's the glibc. But ok, this is not a kernel issue. Darn, so close :)

do a netstat -s on both the sending and receiving boxes to see what the
error counters say. A 1475 byte packet happends to be exactly 1514 bytes
(the mtu max, typically) so you may have an mtu problem

DB

-
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/