Update: [Click] Problem with fast-classifier

[Peter Lippens]

Thanks Giorgio:

- all packet loss is in the queue
- no - packet generator (Smartbit) is directly connected through wire with
Click-machine
- CPU 1800 MHz

I already tried to change all possible parameters, but nothing helps...


Citeren Giorgio Calarco <gcalarco at deis.unibo.it>:

> I (think to) remember that last may or june someone
> was noticing the same using a Simple configuration,
> perhaps it was Nikitas Liogkas ?
> Try to take a look at the list archive to see if you can
> find that thread, maybe you can find any idea
> useful to solve that problem.
>
> In any case...
>
> - where did you registered packet losses,at the queue level
> or at the NIC level ?
> - do you have a wired connection between
> your source and packet switcher or is there
> any layer-2 switch in the middle ?
> - which is the CPU clock frequency ?
>
>
> ciao
> giorgio
>
>
> ----- Original Message -----
> From: "Peter Lippens" <Peter.Lippens at ugent.be>
> To: "Eddie Kohler" <kohler at cs.ucla.edu>
> Cc: <click at amsterdam.lcs.mit.edu>
> Sent: Thursday, November 11, 2004 12:39 PM
> Subject: Re: Update: [Click] Problem with fast-classifier
>
>
> > Hi Eddie,
> >
> > Did anyone, concerning this, had a problem with the normal forwarding
> > througput
> > of Click 1.4.1 without optimizing ?
> > I ask this question because, with an AMD Athlon XP with 2 NIC's Intel
> > Ethernet
> > Pro 100 Mbps (eepro100 standard driver)(one in, other out), and packets of
> > 64
> > byte, and a simple config (PollDevice -Queue - ToDevice)(both polling), I
> > have
> > already packet-loss at +/- 65 Mbps load (+/- 100.000 packets/sec), which
> > is much
> > less than your measurements, and earlier measurements here in Ghent.
> >
> > Thanks in advance,
> >
> > Peter
> >
>
>


--