Re: adsl and altq

[Ralph Kube <ralph_kube_(_at_)_gmx_(_dot_)_net>]

On Mon, Feb 03, 2003 at 03:19:16PM +0100, Henning Brauer wrote:
> 
> well you have to be a bit more specific and give more data ;-)
> you're using -current, e. g. altq merged into pf?
> then you can watch altq working via pfctl -vvsq

Sorry, forgot that. I'm using 3.2-stable. 
The line is 768kBit/s upstream and 128kBit/s downstream. With my config
i try to limit and partition the upstream:

<begin /etc/altq.conf:>

interface tun0 bandwidth 120k cbq 

class cbq tun0 root_class NULL pbandwidth 100

class cbq tun0 bitchtcp root_class pbandwidth 20 maxburst 1 flowvalve
	filter tun0 bitchtcp 0 4661 0 0 6
	filter tun0 bitchtcp 0 0 0 4661 6
	filter tun0 bitchtcp 0 4662 0 0 6
	filter tun0 bitchtcp 0 0 0 4662 6

class cbq tun0 bitchudp root_class pbandwidth 5 maxburst 1 flowvalve
	filter tun0 bitchudp 0 4665 0 0 17
	filter tun0 bitchudp 0 0 0 4665 17
	filter tun0 bitchudp 0 4666 0 0 17
	filter tun0 bitchudp 0 0 0 4666 17
	filter tun0 bitchudp 0 0 0 16283 17

class cbq tun0 ack_class root_class pbandwidth 15 borrow control

class cbq tun0 ssh_class root_class pbandwidth 15
	filter tun0 ssh_class 0 22 0 0 6

class cbq tun0 standard_class root_class pbandwidth 40 borrow maxdelay 10 red default

<end /etc/altq.conf>

I got an application running on another machine on my internal network
which uses a lot of traffic on 4661/4662 tcp and 4665/4666/16283 udp.
To strictly limit its traffic i didn't use the borrow directive. 
Whenever this application is not running i get 'normal' ping times, like
55ms to a good host (www.heise.de).
But whenever it is running my ping goes up to about 2500ms, which makes
it very painful to use other applications, like a browser or a mail
client. 

> it's quite clear some queues are overlimit sometimes; the resolution is
> limited.
This is one of my main concerns. The man page already says that. 
Is altq usable for my situation?

Thank you,
Ralph Kube