Re: Egress bandwidth limiting not working?

[Andrew Beverley <andy@andybev.com>]

On Wed, 2013-05-08 at 19:29 +0100, Sebastian Arcus wrote:
> > The one thing to watch out for is
> > that the sum of the bandwidth for all the leaf classes should add up to
> > the bandwidth of the root class.
> 
> In that case is the HTB example in the LARTC cookbook 
> (http://www.lartc.org/howto/lartc.cookbook.ultimate-tc.html) wrong? It 
> assigns to the first leaf class 100% of parent bandwidth, and to the 
> second 90% of it.

I've seen plenty of examples where people don't seem to have a problem.
However, I found this after a quick search, which seems to make sense:

"Because HTB does not shape in any class except the leaf class, the sum
of the rates of leaf classes should not exceed the ceil of a parent
class. Ideally, the sum of the rates of the children classes would match
the rate of the parent class, allowing the parent class to distribute
leftover bandwidth (ceil - rate) among the children classes"

http://www.tldp.org/HOWTO/Traffic-Control-HOWTO/classful-qdiscs.html

In other words, HTB only looks at each leaf separately when shaping, so
you could potentially have the total of the leaf classes trying to draw
more bandwidth than the total defined.

> It seems that HTB was added to the kernel in 2.4.20! - so it would be 
> nice if somebody with access to LARTC.org would amend the mention about 
> kernel patching on the page above. Also, it would be useful if the same 
> page would mention that HTB is preferable and simpler than CBQ.

Agreed, there is a lot of work to be done updating the documentation.
The LARTC documentation is open to patches though, so if you get a
moment please consider submitting a patch!

Andy