Seems to be a bug, for me the solution was to recompile the kernel.
Under Network/Networking Options/QoS and/or fair queuing you have to set
'Packet scheduler clock source' to 'CPU cycle counter'. You can set this
option for CPUs which have an TSC (e.g. i686), you have to config this
first.
In .config it appears as follows:
# CONFIG_NET_SCH_CLK_JIFFIES is not set
# CONFIG_NET_SCH_CLK_GETTIMEOFDAY is not set
CONFIG_NET_SCH_CLK_CPU=y
Not only SuSE 10.0 is affected, at least 9.3 too.
__
(E-Mail Removed) wrote:
> Hello,
> I'm trying to set up a test network for simulating WAN links but tc
> qdisc netem rules seem to be causing the assigned interface to go nuts
> and die.
>
> I'm running SuSE 10.0 (kernel 2.6.13-15) on a box with two NICs. eth0
> is a 3com 905C, the other interface is an onboard Intel chip. eth0's
> network is masqueraded and forwarded to the other interface.
>
> When I apply the command "tc qdisc add dev eth0 root netem delay 350ms
> loss 1%" everything works beautifully, for a few hundred packets. Pings
> through the interface return the proper delay, everything looks fine.
> However, when I start sending any appreciable test traffic (well below
> 1mbit) through the interface, it stops forwarding traffic. Sometimes it
> dies after 2000 packets, sometimes after 200. I don't seem to be
> hitting a limit in packet count, but could there be a buffer limit
> somewhere I might be hitting?
>
> At any rate, when I execute "tc qdisc del dev eth0 root", the interface
> starts forwarding again.
>
> I've spent a couple of days trying to RTFM, but now I'm putting this
> before the group hoping for some help.
>
> thanks,
> -carl hirsch
Similar Threads
Linear Mode
