--pgp-sign-Multipart_Mon_Jul_25_21:32:48_2005-1
Content-Type: text/plain; charset=US-ASCII

>>>>> "mh" == Martin Husemann <martin@duskware.de> writes:

I switched to hme0 hoping the problem would go away, but it seems to
happen with both hme and gem.

    mh> what does tcpdump -i gem0 say,

# tcpdump -n -p -i hme0
01:23:58.853807 arp who-has 69.31.131.39 tell 69.31.131.33
01:23:59.025760 arp who-has 69.31.131.34 tell 69.31.131.33
01:23:59.031749 arp who-has 69.31.131.61 tell 69.31.131.33
01:23:59.677599 arp who-has 69.31.131.42 tell 69.31.131.33
01:23:59.700073 802.1d config 8000.00:0b:db:fc:74:08.800c root 8000.00:01:62:04:0a:60 pathcost 100008 age 3 max 20 hello 2 fdelay 15 
01:24:00.026512 arp who-has 69.31.131.34 tell 69.31.131.33
01:24:00.032518 arp who-has 69.31.131.61 tell 69.31.131.33
01:24:00.171488 arp who-has 69.31.131.39 tell 69.31.131.33
01:24:01.026287 arp who-has 69.31.131.34 tell 69.31.131.33
01:24:01.032272 arp who-has 69.31.131.61 tell 69.31.131.33
01:24:01.144245 arp who-has 69.31.131.42 tell 69.31.131.33
01:24:01.171233 arp who-has 69.31.131.39 tell 69.31.131.33
01:24:01.739251 802.1d config 8000.00:0b:db:fc:74:08.800c root 8000.00:01:62:04:0a:60 pathcost 100008 age 3 max 20 hello 2 fdelay 15 
01:24:02.082020 arp who-has 69.31.131.61 tell 69.31.131.33
01:24:02.143996 arp who-has 69.31.131.42 tell 69.31.131.33
01:24:02.170990 arp who-has 69.31.131.39 tell 69.31.131.33
01:24:03.081776 arp who-has 69.31.131.61 tell 69.31.131.33
01:24:03.143764 arp who-has 69.31.131.42 tell 69.31.131.33
01:24:03.195764 arp who-has 69.31.131.39 tell 69.31.131.33
01:24:03.698476 802.1d config 8000.00:0b:db:fc:74:08.800c root 8000.00:01:62:04:0a:60 pathcost 100008 age 3 max 20 hello 2 fdelay 15 

69.31.131.34 is the IP assigned to hme0.  The other hosts mentioned in
the 'arp who-has' are ones I am proxy-arping for with 'choparp' and
then forwarding traffic to them over a gre tunnel.  69.31.131.33 is the 
default gateway.

    mh> does the gem still interrupt

a little.  I think it is 'lev6' because if I 'ifconfig hme0 down' then
the lev6 row stops incrementing.  gem1 is ifconfigged up, but not
plugged in.

# vmstat -i
interrupt                                     total     rate
intr lev1                                  12723676      360
intr lev4                                     18690        0
intr lev5                                    306659        8
intr lev6                                  22571054      639
intr clock                                  3530919       99
intr lev12                                    18755        0
gem1 interrupts                                  11        0
Total                                      39169764     1109
# sleep 5
# vmstat -i
interrupt                                     total     rate
intr lev1                                  12724203      360
intr lev4                                     18690        0
intr lev5                                    306686        8
intr lev6                                  22571074      639
intr clock                                  3531422       99
intr lev12                                    18755        0
gem1 interrupts                                  11        0
Total                                      39170841     1109

    mh> does resetting altq help (instead of gem0 down/up).

ALTQ is not compiled into this NetBSD 3.0 kernel.

--pgp-sign-Multipart_Mon_Jul_25_21:32:48_2005-1
Content-Type: application/pgp-signature
Content-Transfer-Encoding: 7bit

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (NetBSD)

iQCVAwUAQuWSwInCBbTaW/4dAQKXWQP/bp1zdPwQjhFZ5oYBQpZfcbkQ0xvg0aRQ
j7RXYwhF/a7SH2PhOBtqW0kDydu5uQjCCZDY6pZC884dmNpoGmHCSTsiaCqiiDLT
WYOKEJHcFsE0cVV1dnqpH5iY1NO6/JChct4RYLmSiAXGYj8rTPJkNM+9/W2J7lwV
D2T4vfftI0U=
=teDf
-----END PGP SIGNATURE-----

--pgp-sign-Multipart_Mon_Jul_25_21:32:48_2005-1--