Re: Packet loss with traffic shaper and routing
- From: Lee Johnston <lee@xxxxxxxxxxxxxxx>
- Date: Wed, 26 Apr 2006 16:15:21 +0100
Hi,
Try using device polling to reduce the number of interrupts.
Add this to your kernel:
options DEVICE_POLLING
options HZ=1000
And set sysctl kern.polling.enable=1
Hope this helps,
Regards,
Lee.
tpeixoto@xxxxxxxxxxxxxxx wrote:
Hello all!
We have a machine working as a router and bandwidth limiter for our
network. It routes the traffic through two 'bge' interfaces utilizing only
public IPs. No NAT is used. We only have IPFW rules for traffic shaping by
MAC addresses.
It works fine, but we have been experiencing high latency and packet loss.
There is no other major services in this machine, cpu utilization is low,
memory is fine, no disk activity, but load average is always around 1.5 or
2, even if I disable IPFW layer 2 filtering.
Network traffic is not greater than 12 Mbit/s.
The system is a FreeBSD 5.4-RELEASE, running on a dual CPU: Intel(R)
Xeon(TM) CPU 3.06GHz (3051.47-MHz 686-class CPU) compiled with SMP kernel:
cpu0 (BSP): APIC ID: 0
cpu1 (AP): APIC ID: 1
cpu2 (AP): APIC ID: 6
cpu3 (AP): APIC ID: 7
top output:
last pid: 14039; load averages: 1.61, 1.62, 1.56
37 processes: 1 running, 35 sleeping, 1 zombie
CPU states: 0.0% user, 0.0% nice, 1.4% system, 30.8% interrupt, 67.8% idle
Mem: 16M Active, 421M Inact, 149M Wired, 688K Cache, 112M Buf, 1417M Free
Swap: 1024M Total, 1024M Free
PID USERNAME PRI NICE SIZE RES STATE C TIME WCPU CPU COMMAND
6245 root 96 0 2672K 2236K select 2 3:48 0.00% 0.00% dhcpd
329 root 96 0 1328K 892K select 0 2:48 0.00% 0.00% syslogd
453 root 96 0 3384K 2508K select 2 0:33 0.00% 0.00% sshd
systat -iostat output:
/0 /1 /2 /3 /4 /5 /6 /7 /8 /9 /10
Load Average ||||||||||
/0 /10 /20 /30 /40 /50 /60 /70 /80 /90 /100
cpu user|
nice|
system|
interrupt|XXXXXXXXXXXXXXXXX
idle|XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
/0 /10 /20 /30 /40 /50 /60 /70 /80 /90 /100
da0 MB/s
tps|X
pass0 MB/s
tps|
Can anyone give me some advice?
Thank you in advance!
_______________________________________________
freebsd-net@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-net
To unsubscribe, send any mail to "freebsd-net-unsubscribe@xxxxxxxxxxx"
--
------------------------------------------------------------------------
Wildcard Internet - end-to-end internet solutions
Lee Johnston - Wildcard Internet
Tel: 0845 165 1510
Fax: 0845 165 1511
Email: lee@xxxxxxxxxxxxxxx <mailto:lee@xxxxxxxxxxxxxxx>
Web: www.wildcard.net.uk <http://www.wildcard.net.uk>
_______________________________________________
freebsd-net@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-net
To unsubscribe, send any mail to "freebsd-net-unsubscribe@xxxxxxxxxxx"
- Follow-Ups:
- Re: Packet loss with traffic shaper and routing
- From: tpeixoto
- Re: Packet loss with traffic shaper and routing
- References:
- Packet loss with traffic shaper and routing
- From: tpeixoto
- Packet loss with traffic shaper and routing
- Prev by Date: Packet loss with traffic shaper and routing
- Next by Date: Re: Packet loss with traffic shaper and routing
- Previous by thread: Packet loss with traffic shaper and routing
- Next by thread: Re: Packet loss with traffic shaper and routing
- Index(es):
Relevant Pages
- Oops with 2.6.1_rc1-mm1
... Unable to handle kernel NULL pointer dereference at virtual address ... CPU:
0 ... EIP is at vt_ioctl+0x1e/0x1f00 ... Pin 2-17 already programmed ...
(Linux-Kernel) - [OOPS] hotplugging cpus via /sys/devices/system/cpu/
... then the CPU doesn't start and the kernel prints out an oops. ... Using
1 I/O APICs ... Linux Plug and Play Support v0.97 Adam Belay ... (Linux-Kernel) - (ondemand) CPU governor regression between 2.6.23 and 2.6.24
... With the ondemand governor the CPU runs at 600 MHz, ... The ondeman governor
of previous kernel versions however automatically increased ... # PCI IDE chipsets support
... (Linux-Kernel) - Clocksource tsc is always unstable with 2.6.25-* kernels and CONFIG_NO_HZ=y on my box
... Switched to high resolution mode on CPU 1 ... Another strange thing is
when I try to boot that kernel with clocksource=acpi_pm it just hangs. ... # Parallel
IDE high-level drivers ... (Linux-Kernel) - i386 smp (on pentium IV with HT) kernel (2.4.26) crash at sched.c:681
... It is a P4 with HT running a 2.4.26 smp kernel that I ... IntelPentium4 CPU
2.80GHz ... cache size: 512 KB ... PCI: PCI BIOS revision 2.10 entry at 0xfba63,
... (Linux-Kernel)