Re: [PATCH] Improve behaviour of Netlink Sockets

[jamal <hadi@xxxxxxxxxx>]

On Thu, 2004-09-23 at 08:05, Herbert Xu wrote:

> > Try explicitly reducing the socket buffer size and see what happens
> 
> What's wrong with the default sizes? If you decrease it far enough
> of course it's going to overflow.

The idea is to reproduce the overun ;->

> > set i to 10000 or a little more
> 
> You might want to optimise your add path.  It was painful with 20000 :)

Probably debug prints slowing things; you could of course do some
batching like so:

$TC actions add \
 action drop index 1
 action drop index 2 
 action drop index 3 ...

Do 10 a time in the for loop and should be roughly 10 times as fast

> > to dump: tc actions ls action drop
> 
> That was much faster, no overflows at all:

I apologize i dont have time right now to chase it and experiment; my
test machine can be made to reproduce it. Maybe this weekend

cheers,
jamal