Alexey's assesment that the culprit might be he might be loosing some
of those frags in the backlog
What kind of hardware is this thing running on? 1.5 G seems a lot on
a regular PC. very Large MTUs?
Bob, One of the things that could be done is have the driver stop sending
packets after the toplayer gets congested (return is NET_RX_CN_HIGH).
linux/include/linux/netdevice.h has the NET_RX_* which are returned
by netif_rx()
cheers,
jamal
On Fri, 9 Mar 2001, Andi Kleen wrote:
> On Thu, Mar 08, 2001 at 07:56:43PM +0100, Bob Felderman wrote:
> > I think the culprit here is the packet receive errors from
> > UDP. These are socket overflows I think. I'm trying to
> > track it down. When I used DaveM's zero-copy patches on
> > a linux-2.4.0 kernel, most, if not all, of these
> > packet receive errors went away.
>
> So does it help when you increase the socket receive buffer of the receiver ?
>
> (via /proc/sys/net/core/rmem_{default,max})
>
>
> -Andi
> >
>
|