Subject: More fun with ARP tables
To: None <port-sparc@netbsd.org>
From: Greg Earle <earle@isolar.Tujunga.CA.US>
List: port-sparc
Date: 11/02/1994 12:47:16
One for the "kids, don't try this at home" dept:

I'd done a "rup" and had a pretty populated ARP table as a result (52 entries).

All the logical link entries were cluttering up my routing tables, so I
manually trashed a bunch of 'em via "arp -d".  A few minutes later, I did an
"arp -a" with the result being

netbsd4me.jpl.nasa.gov:1:38 % arp -a
b168-fddi (128.149.24.1) at aa:0:4:0:43:38
elroy.jpl.nasa.gov (128.149.24.3) at 8:0:20:8:ac:1f
scn9.jpl.nasa.gov (128.149.24.10) at 8:0:14:20:38:60
mipl7.jpl.nasa.gov (128.149.24.11) at 0:0:0:3:40:1
                                      ^^^^^^^^^^^^
tone.jpl.nasa.gov (128.149.24.30) at (incomplete)
                                     ^^^^^^^^^^^^
macgw1180.jpl.nasa.gov (128.149.24.45) at 0:0:18:0:7:24
andes.jpl.nasa.gov (128.149.24.108) at 8:0:69:7:bb:10
teton.jpl.nasa.gov (128.149.24.109) at 0:0:0:3:40:1
                                       ^^^^^^^^^^^^
centinela.jpl.nasa.gov (128.149.24.123) at 8:0:20:12:53:c5
alps.jpl.nasa.gov (128.149.24.126) at 0:0:0:3:40:1
                                      ^^^^^^^^^^^^
temblor.jpl.nasa.gov (128.149.24.127) at 8:0:20:17:f9:e4
wasatch.jpl.nasa.gov (128.149.24.128) at 0:89:80:3:40:1
                                         ^^^^^^^^^^^^^^
hermes.jpl.nasa.gov (128.149.24.165) at 8:0:69:2:35:c6
isolar.tujunga.ca.us (128.149.24.254) at 8:0:20:9:4a:7

All of the above tagged entries are trashed.

Another minute or so later and more trashed/changed entries exist:

mipl7.jpl.nasa.gov (128.149.24.11) at 0:0:0:2:0:0
teton.jpl.nasa.gov (128.149.24.109) at 0:0:0:2:0:0
wasatch.jpl.nasa.gov (128.149.24.128) at 0:89:80:2:0:0
alps.jpl.nasa.gov (128.149.24.126) at 0:0:0:2:0:0

Somewhat unsurprisingly, telnet to these hosts failed, with the error

Nov  2 10:48:13 netbsd4me /netbsd: arptnew failed on 8095187e
Nov  2 10:48:42 netbsd4me last message repeated 2 times
Nov  2 10:49:48 netbsd4me /netbsd: arptnew failed on 8095180b

About 6 minutes later things were working again, presumably after the
original entry timed out.

Anyway, just thought I'd mention that.  Seems to be a fairly easy way to
make the problem occur.

Of course, I'm still using the now-3-month-old (hint hint (-:) binary snapshot
for userland, ... but I doubt it's been fixed, because I just built a new
kernel (with up-to-date sup kernel source) a few days ago.

BTW, X11 R6 works just peachy on a GX!!!  Nice going, folks.

Now if only I could use a serial port at higher than 9600 baud (lots and
lots of "zs0a: fifo overrun" errors at 19200 and 38400; "tip" drops lots of
characters ... )

	- Greg