On Thu, 24 Jun 1999, Erik Andersen wrote:
> On Wed, Jun 23, 1999 at 08:16:40PM -0400, jamal wrote:
> >
> >
> > > Do you want the output of the above mentioned 'tc' commands now (while
> > > it is working) or after an hour or so (when it stops)?
> >
> > Can you post your script, output etc and point out exactly which interface
> > it happens on?
> >
>
> r2d2:~# tc -s -d qdisc ls dev eth0
> qdisc cbq 10: rate 100Mbit cell 8b (bounded,isolated) prio no-transmit/8 weight
>100Mbit allot 1514b
> level 2 ewma 5 avpkt 1000b maxidle 1us
Check dmesg output immediately after you set up. As a short term remedy,
also try to set the weight to 1 (both in the WEIGHT and DEVICE params) and
repeat.
It seems to me the problem could be related to the small rates you set.
Could you try to increase the rates and the minimal being 1mbit (from
128kbit and 2mbit from 256kbit) and see if this problem is still there?
Also from the quick scan of that script i think the author is still using
the old u32 syntax. The port number probably never works (since there is
no specification whether it is TCP or UDP). I could be wrong.
> r2d2:~# tc -s -d qdisc ls dev eth2
> qdisc tbf 803e: rate 128Kbit burst 10Kb/8 mpu 0b lat 1.7s
> Sent 92578 bytes 817 pkts (dropped 0, overlimits 0)
> backlog 21041b 294p
That is definetly a huge backlog of small packets (looks like mostly
SYNs or pings).
cheers,
jamal
PS:
I'll try and see if i can reproduce your problme in my setup after you
test with the above.
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to [EMAIL PROTECTED]
