* [LARTC] tc troubles
@ 2000-10-23 10:27 Hoomaan
2000-10-23 11:24 ` Hoomaan
0 siblings, 1 reply; 2+ messages in thread
From: Hoomaan @ 2000-10-23 10:27 UTC (permalink / raw)
To: lartc
<PRE>Hi all,
Its a while we are trying to use "tc" for bandwidth control. But each time
we apply tc rules to the network, we experience a lot of service
degradation. The usage we are mostly concerned about, is bandwidth control
of a group of IP addresses. We mark that subnet with IPChains and then apply
the tc rules. These are the rules applied to eth5 in order to limit the
uplink of packets marked "7" by ipchains:
# /sbin/tc class sh dev eth5
class cbq 40: root rate 100Mbit (bounded,isolated) prio no-transmit
class cbq 40:1 parent 40: rate 100Mbit prio no-transmit
class cbq 40:300 parent 40:1 leaf 8025: rate 200Kbit (bounded) prio 5
# /sbin/tc qdisc sh dev eth5
qdisc tbf 8025: rate 200Kbit burst 25Kb lat 1us
qdisc cbq 40: rate 100Mbit (bounded,isolated) prio no-transmit
[<A HREF="mailto:root@qos-1">root@qos-1</A> naimi]# /sbin/tc filter sh dev eth5
filter parent 40: protocol ip pref 1 fw
filter parent 40: protocol ip pref 1 fw handle 0x7 classid 40:300
filter parent 40: protocol ip pref 80 u32
filter parent 40: protocol ip pref 80 u32 fh 800: ht divisor 1
filter parent 40: protocol ip pref 80 u32 fh 800::800 order 2048 key ht 800
bkt 0 flowid 40:100
As I mentioned, the service degraded severely: pig time of internet sites
increased about 300-500 ms!!
Does anybody know what is wrong with my configuration?
Best Regards
Hoomaan Naimi
Afranet Network Administrator
</PRE>
^ permalink raw reply [flat|nested] 2+ messages in thread
* [LARTC] tc troubles
2000-10-23 10:27 [LARTC] tc troubles Hoomaan
@ 2000-10-23 11:24 ` Hoomaan
0 siblings, 0 replies; 2+ messages in thread
From: Hoomaan @ 2000-10-23 11:24 UTC (permalink / raw)
To: lartc
<PRE>Dear Sir,
The question exactly arises from here:
How can I monitor the impact of these several settings on the network
performance?
How can I monitor the size of the queues, log the packets dropped ,....?
I would be thankful if you recommend me some tools or APIs for doing these
tasks.
Best Regards
Hoomaan Naimi
Afranet Network Administrator
----- Original Message -----
From: "bert hubert" <<A HREF="mailto:ahu@ds9a.nl">ahu@ds9a.nl</A>>
To: "Hoomaan Naimi" <<A HREF="mailto:naimi@afranet.com">naimi@afranet.com</A>>
Sent: Monday, October 23, 2000 1:59 PM
Subject: Re: [LARTC] tc troubles
><i> On Mon, Oct 23, 2000 at 01:45:22PM +0330, Hoomaan Naimi wrote:
</I>><i> > Hi all,
</I>><i> > Its a while we are trying to use "tc" for bandwidth control. But each
</I>time
><i> > we apply tc rules to the network, we experience a lot of service
</I>><i> > degradation. The usage we are mostly concerned about, is bandwidth
</I>control
><i> > of a group of IP addresses. We mark that subnet with IPChains and then
</I>apply
><i> > the tc rules. These are the rules applied to eth5 in order to limit the
</I>><i> > uplink of packets marked "7" by ipchains:
</I>><i>
</I>><i> Try reading the source of CBQ.init, which can set up rules for you. I miss
</I>a
><i> lot of settings in your output, like weight, cell size etcetera. This
</I>might
><i> lead to a bad configuration.
</I>><i>
</I>><i> Regards,
</I>><i>
</I>><i> bert hubert
</I>><i>
</I>><i> > # /sbin/tc class sh dev eth5
</I>><i> > class cbq 40: root rate 100Mbit (bounded,isolated) prio no-transmit
</I>><i> > class cbq 40:1 parent 40: rate 100Mbit prio no-transmit
</I>><i> > class cbq 40:300 parent 40:1 leaf 8025: rate 200Kbit (bounded) prio 5
</I>><i> > # /sbin/tc qdisc sh dev eth5
</I>><i> > qdisc tbf 8025: rate 200Kbit burst 25Kb lat 1us
</I>><i> > qdisc cbq 40: rate 100Mbit (bounded,isolated) prio no-transmit
</I>><i> > [<A HREF="mailto:root@qos-1">root@qos-1</A> naimi]# /sbin/tc filter sh dev eth5
</I>><i> > filter parent 40: protocol ip pref 1 fw
</I>><i> > filter parent 40: protocol ip pref 1 fw handle 0x7 classid 40:300
</I>><i> > filter parent 40: protocol ip pref 80 u32
</I>><i> > filter parent 40: protocol ip pref 80 u32 fh 800: ht divisor 1
</I>><i> > filter parent 40: protocol ip pref 80 u32 fh 800::800 order 2048 key ht
</I>800
><i> > bkt 0 flowid 40:100
</I>><i> >
</I>><i> > As I mentioned, the service degraded severely: pig time of internet
</I>sites
><i> > increased about 300-500 ms!!
</I>><i> > Does anybody know what is wrong with my configuration?
</I>><i> >
</I>><i> > Best Regards
</I>><i> > Hoomaan Naimi
</I>><i> > Afranet Network Administrator
</I>><i> >
</I>><i> >
</I>><i> >
</I>><i> >
</I>><i> > _______________________________________________
</I>><i> > LARTC mailing list / <A HREF="mailto:LARTC@mailman.ds9a.nl">LARTC@mailman.ds9a.nl</A>
</I>><i> > <A HREF="http://mailman.ds9a.nl/mailman/listinfo/lartc">http://mailman.ds9a.nl/mailman/listinfo/lartc</A> HOWTO:
</I><A HREF="http://ds9a.nl/2.4Routing/">http://ds9a.nl/2.4Routing/</A>
><i> >
</I>><i>
</I>><i> --
</I>><i> PowerDNS Versatile DNS Services
</I>><i> Trilab The Technology People
</I>><i> 'SYN! .. SYN|ACK! .. ACK!' - the mating call of the internet
</I>
</PRE>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2000-10-23 11:24 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2000-10-23 10:27 [LARTC] tc troubles Hoomaan
2000-10-23 11:24 ` Hoomaan
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.