[BUGS] freebsd ethernet traffic load

[jonathan michaels]

greetings all,

-----Forwarded message from Charlie Root <@gw>-----
Network interface status:
Name    Mtu Network       Address              Ipkts Ierrs    Opkts Oerrs  Coll
de0    1500 <Link#1>      00:00:e8:XX:XX:XX  4593322     0  4779269     0 179288 
de0    1500 caamora-au/28 gw                 2202190     -  3140772     -     - 
plip0  1500 <Link#2>                               0     0        0     0     0 
lo0   16384 <Link#3>                           16623     0    16623     0     0 
lo0   16384 fe80:3::1     fe80:3::1                0     -        0     -     - 
lo0   16384 localhost.caa ::1                      4     -        4     -     - 
lo0   16384 127           localhost            12115     -    12115     -     - 
pflog 33208 <Link#4>                               0     0        0     0     0 
tun0   1500 <Link#5>                         4343502     0  2933153     0     0 
tun0   1500 XXX,XX,XX,XXX XXXXXXX.XXX.XXXXX   744520     -  1246901     -     - 

Local network system status:
host_1        up  28+11:31,     0 users,  load 0.00, 0.00, 0.00
host_gw       up  86+05:29,     0 users,  load 0.00, 0.00, 0.00
host_2        up  35+23:45,     0 users,  load 0.00, 0.00, 0.00
host_3        up   6+17:21,     0 users,  load 0.00, 0.00, 0.02
-----End of forwarded message-----

i have been seeing some interestings over teh last couple of
months (6 more or less). the network traffic is changing.

bit of background, the hardware is pretty old (but reilable and
stable, as well, i've had it since new) it was also top shelf at
teh time.

_gw has recently started loging "Coll" on de0 

de0: <Digital 21040 Ethernet> port 0x6100-0x617f mem 0xe4001000-0xe400107f irq 9 at device 11.0 on pci0
de0: ACCTON EN1203 21040 [10Mb/s] pass 2.4

this host is teh network gateway, it is running a recentversion
of freebsd (upgraded from v2.2.7-RELEASE about march/april 2007)

FreeBSD 6.2-RELEASE #0: Fri Jan 12 10:40:27 UTC 2007

CPU: Pentium/P54C (132.87-MHz 586-class CPU)
  Origin = "GenuineIntel"  Id = 0x52c  Stepping = 12
  Features=0x1bf<FPU,VME,DE,PSE,TSC,MSR,MCE,CX8>
real memory  = 67108864 (64 MB)
avail memory = 56098816 (53 MB)

all of its existance it has been teh network gateway with the
same nic and analogue dialup modem running freebsd kernel pppd
(but in 6.2 i shifted to userland ppp, the plan is to move to
mpd as soon as i can workout how to configure mpd (sugestions
appreciated) on a netcomm roadster or a usr v.everything both
running at 33k6 bits/s. these collisions have only started
since the rest of teh network was upgraded to 100 mhz
operating (switch 10/100 mhz telesys "centrecom fs708").

these collisions only started when i moved teh rest of teh
network to 100 mhz and specifically a 4 cpu machine with a
intel 10/100 mhz nic (proliant 6400r 4 gb/4x500mhz etc scsi
raid and 10/100 intel ethernet).

also, i am planing to replace the switch (fs708) with one of
teh recently aquired switchs, be it teh d-link ?? or a
cabletron ssr2000 or the other one (much appreciations edwin).
i don't think the swtich is teh problem, well, its only an 8
port beastie and thats a bit small at teh moment .. <GRIN>

could it be possible that the host is having trouble keeping up
with teh workload especially now that it is running pf (as teh
firewall, etc) and x11 for a gui interface (which i can ditch).

i'm looking fro informed comment before i spen anymore on this
machine, especially in terms off maney. particularly given that
sooner than latter this machine is going to have some sort of
adsl modem/router pushing several hundred orders of magnitude
of traffic .. do i need to look seriously at replacing teh
machine, is it the 'weakest' link ?? in this chain ?

most kind regards and appreications

jonathan

-- 
================================================================
powered by ..
QNX, OS9 and freeBSD  --  http://caamora com au/operating system
==== === appropriate solution in an inappropriate world === ====