Skip to contentSkip to conversation

dsl latency...

LARTC Mattt 2 participants 3 messages Oct 4, 2002

lore
Share & tools
lore

Mattt

author
#0 lore raw

Hi all,

I've gotten the WonderShaper (slightly modified) running on the
router. We have a 512/128 connection, and I set DOWNLINK00, UPLINK0 (the link is currently under-utilised, and huge performance is not required at this stage). All appears well (at least, it's stable), but I have two questions.

Firstly, given that the link is *very* under-utilised, does the
following output look reasonable? It looks as if only qdisc has actually seen traffic :

=====

jenner:/etc/firewall/wondershaper-1.1a# ./wshaper status
qdisc ingress ffff: ---------------- 
 Sent 0 bytes 0 pkts (dropped 0, overlimits 0) 

qdisc sfq 30: quantum 1514b perturb 10sec
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)

qdisc sfq 20: quantum 1514b perturb 10sec
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)

qdisc sfq 10: quantum 1514b perturb 10sec
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)

qdisc cbq 1: rate 10Mbit (bounded,isolated) prio no-transmit
Sent 2896610 bytes 29310 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0

class cbq 1: root rate 10Mbit (bounded,isolated) prio no-transmit
Sent 2896610 bytes 29310 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0
class cbq 1:10 parent 1:1 leaf 10: rate 100Kbit prio 1
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0
class cbq 1:1 parent 1: rate 100Kbit (bounded,isolated) prio 5
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0
class cbq 1:20 parent 1:1 leaf 20: rate 90Kbit prio 2
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0
class cbq 1:30 parent 1:1 leaf 30: rate 80Kbit prio 2
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0
jenner:/etc/firewall/wondershaper-1.1a#
=====

Also, I'm noticing a fair wait before, for instance, downloading a web
site (although the phenomenom is also quite apparent over at least most protocols, though). Say, perhaps, 1 or 2 seconds before *any* page 'instantaneously' appears ;-)

We run our own DNS, as well as a DNS cache (the djbdns package), so
lookups should not be causing a problem (in fact, they're not - this only happens to traffic leaving the DSL interface).

I realise that DSL latency isn't as good as some other technologies,
but is this something I should be able to minimise the effect of?

Admittedly, it sounds to me as if the traffic is still queuing at the
modem - have I simply done something stupid? I'm applying the qdiscs to eth3 rather than ppp0 (as applying them to ppp0 would oops the kernel after less than a minute or two reliably...).

I'm new to lartc, but learning (through necessity). I can't help but
feel that my questions here are actually related - the lack of counter data on the qdiscs, the classic symtoms(?) of DSL latency... Is it even working for me?

Signature · 10 lines
Cheers,
 Mattt.                      icq   : 117539757 
 aboveNetworks               www   : www.above.nq4u.net
 [email protected]        jabber: [email protected]
     
	 What's got four legs and an arm?  A happy Pit Bull...

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
author
#1 lore raw

Was just reading over the script, and remembered that I re-installed it fresh... it's *not* slightly modified as stated below, other than the d/u link numbers... the noprio stuff is all empty...

Quoted 82 lines · Mattt

Hi all,

I've gotten the WonderShaper (slightly modified) running on the
router. We have a 512/128 connection, and I set DOWNLINK00, UPLINK0 (the link is currently under-utilised, and huge performance is not required at this stage). All appears well (at least, it's stable), but I have two questions.

Firstly, given that the link is *very* under-utilised, does the
following output look reasonable? It looks as if only qdisc has actually seen traffic :

=====

jenner:/etc/firewall/wondershaper-1.1a# ./wshaper status
qdisc ingress ffff: ---------------- 
 Sent 0 bytes 0 pkts (dropped 0, overlimits 0) 

qdisc sfq 30: quantum 1514b perturb 10sec
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)

qdisc sfq 20: quantum 1514b perturb 10sec
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)

qdisc sfq 10: quantum 1514b perturb 10sec
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)

qdisc cbq 1: rate 10Mbit (bounded,isolated) prio no-transmit
Sent 2896610 bytes 29310 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0

class cbq 1: root rate 10Mbit (bounded,isolated) prio no-transmit
Sent 2896610 bytes 29310 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0
class cbq 1:10 parent 1:1 leaf 10: rate 100Kbit prio 1
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0
class cbq 1:1 parent 1: rate 100Kbit (bounded,isolated) prio 5
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0
class cbq 1:20 parent 1:1 leaf 20: rate 90Kbit prio 2
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0
class cbq 1:30 parent 1:1 leaf 30: rate 80Kbit prio 2
Sent 0 bytes 0 pkts (dropped 0, overlimits 0)
borrowed 0 overactions 0 avgidle 624 undertime 0
jenner:/etc/firewall/wondershaper-1.1a#
=====

Also, I'm noticing a fair wait before, for instance, downloading a web
site (although the phenomenom is also quite apparent over at least most protocols, though). Say, perhaps, 1 or 2 seconds before *any* page 'instantaneously' appears ;-)

We run our own DNS, as well as a DNS cache (the djbdns package), so
lookups should not be causing a problem (in fact, they're not - this only happens to traffic leaving the DSL interface).

I realise that DSL latency isn't as good as some other technologies,
but is this something I should be able to minimise the effect of?

Admittedly, it sounds to me as if the traffic is still queuing at the
modem - have I simply done something stupid? I'm applying the qdiscs to eth3 rather than ppp0 (as applying them to ppp0 would oops the kernel after less than a minute or two reliably...).

I'm new to lartc, but learning (through necessity). I can't help but
feel that my questions here are actually related - the lack of counter data on the qdiscs, the classic symtoms(?) of DSL latency... Is it even working for me?

Signature · 10 lines
Cheers,
 Mattt.                      icq   : 117539757 
 aboveNetworks               www   : www.above.nq4u.net
 [email protected]        jabber: [email protected]
     
	 What's got four legs and an arm?  A happy Pit Bull...

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
Signature · 10 lines
Cheers,
 Mattt.                      icq   : 117539757 
 aboveNetworks               www   : www.above.nq4u.net
 [email protected]        jabber: [email protected]
     
	 What's got four legs and an arm?  A happy Pit Bull...

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
#2 lore raw
In reply to Mattt
Quoted 3 lines · Mattt

Was just reading over the script, and remembered that I re-installed it fresh... it's *not* slightly modified as stated below, other than the d/u link numbers... the noprio stuff is all empty...

All packets are ending up in the root qdisc so you have a problem with your filters. Can you check it out ?
With no filters you are not shaping at all .....

Stef

Signature · 8 lines
[email protected]
 "Using Linux as bandwidth manager"
     http://www.docum.org/
     #lartc @ irc.oftc.net

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/