[ltp] T60 and Network

[Thomas Finneid]

Jerome Poggi wrote:
> Hi,
> 
> I recently buy a T60 and I install a Gentoo in this new PC.
> I have a problem with the network, just look to a ping :
> # hping -S -p 22 192.168.106.123
> HPING 192.168.106.123 (eth0 192.168.106.123): S set, 40 headers + 0 data
> bytes
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=0 win=5840 rtt=965.5 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=1 win=5840 rtt=1004.9 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=2 win=5840 rtt=0.9 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=3 win=5840 rtt=1004.7 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=4 win=5840 rtt=0.7 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=5 win=5840 rtt=971.8 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=6 win=5840 rtt=943.7 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=7 win=5840 rtt=1007.1 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=8 win=5840 rtt=0.9 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=9 win=5840 rtt=1004.8 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=10 win=5840 rtt=0.8 ms
> len=46 ip=192.168.106.123 ttl=140 DF id=0 sport=22 flags=SA seq=11 win=5840 rtt=431.7 ms

I have earlier also noticed this network jitter problem, but now I have 
made a discovery about something that affects it. Here is a description 
of the symptoms


The output is that every second ping response has a time of about 
1000ms, while the others responses have a normal time of about 20-500ms.
I have tested from other networks and the problem persists. I have also 
tested it with other computers on the same networks, and then there is 
no problem.

By accident I noticed that if the hardisk is doing some working, as in 
full throttle work, the ping response times drops to the normal 20-40 
ms, as soon as the harddisk stops doing any work I experience the 
network jitter again. I dont know if its the fact that the disk is 
working or if its because of the processor having higher load (havent 
found a suitable cpu test to confirm/contradict that last part yet) It 
might also be because of some device io, or a timerinterrupt which is 
disabled during device io.

The machine is a thinkpad t60, with intel core duo, and a sata disk (run 
by the scsi system, not libsata). The kernel is a homemade 2.6.17.8 smp 
with preemtpive locking and a 1000hz clock. (Basically its optimised for 
desktop performance, not server performance. The problem has also 
existed on a previous kernel I used, 2.6.16.9)

The first thing of course is actually trying to identify what it is that 
causes this erratic behaviour. So the question is, does anybody have any 
ideas where the problem might be or have any qualified suggestions as to 
what I should look at.

thomas