SUMMARY : bad lnth packets

Date: Fri Jul 31 1992 - 02:55:09 CDT

Once again, the answers came before I saw my question. You remember,
I asked about messages like "bad lnth 42 dst ffffffffffff src 08002008f47f"
from rpc.etherd. And I forgot to mention my configuration, which seems to
be irrelevant anyways. And I mentioned a PD tool called ethertop, which
is something like traffic or xtr for a dumb terminal (and I like it).

Rohit.Aggarwal@Corp.Sun.COM says :

> This is due to loopback packets that are sent to the NIT interface.
> These packets are less then 60 bytes that ethernet wants. This is
> a bug in the ethernet driver where it loops the packet back to
> the nit interface before padding it. There should be a patch
> available for this shortly.
> There is NO hardware problem. For now I would ignore this message.

Chris Keane <chris@rufus.state.COM.AU> says :

> Naaa, no worries. All it's saying is that you occasionally send out
> "runt" packets (minimum ethernet length is 60). An ICMP or ARP message is
> likely to be about 42 bytes (this is the "real" length of a ping request,
> for example).
> I believe rpc.etherd sits between the IP layer and the ethernet driver.
> The padding from 42 to 60 bytes occurs in the ethernet driver and therefore
> rpc.etherd sees this packet coming out of the IP layer that is a runt and
> reports it as such.
> The real worry is when you get this message from *another* machine since it
> means that the other machine is spewing runts onto the network and not
> padding them. (Steve Jay {Ultra Unix SW Mgr}) says :

> It's not a hardware problem. The message is caused by a bug in a Sun
> kernel routine. It happens whenever the host sends out an arp request,
> which is incorrectly padded when it's forwarded to rpc.etherd. The
> message is harmless, but it's annoying. We reported the bug to Sun many
> months ago, and they just this week sent us a patch to try out. I don't
> think it has an official patch number yet.
> I wasn't the one here who dealt with Sun on this problem, so I don't
> have the bug number. Tomorrow, I'll get the bug number and send it to
> you, and you can use it to request the fix from Sun. (Mark C. Henderson) agrees :

> Probably arp packets. They really go out on the ethernet with length
> 60 (they're padded to that length), but it appears to the local etherd
> that they have length 42. Don't worry about it.

Thanks a lot to you all.

Eckhard R"uggeberg

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:46 CDT