LARTC - Feb 2006 - htb root don''t reach ceil rate?

Hello,

i have a htb setup where the root and (nearly) all child classes has a 
ceil rate with max up from my adsl line. But the root class don''t reach
the ceil value but some childs are get a huge backlog.

My setup: (tc -d class show dev ppp0) [cleaned a bit]
********************************************************
class htb 1:1 root rate 576000bit ceil 576000bit burst 30Kb/8 cburst 
1739b/8 overhead 14b level 7
class htb 1:10 parent 1:1 leaf 100: prio 0 quantum 7500 rate 58000bit 
ceil 115000bit burst 1480b/8 cburst 1508b/8 overhead 14b level 0
class htb 1:20 parent 1:1 leaf 200: prio 1 quantum 256 rate 282000bit 
ceil 576000bit burst 396b/2 cburst 543b/2 overhead 14b level 0
class htb 1:30 parent 1:1 leaf 300: prio 2 quantum 9000 rate 117000bit 
ceil 576000bit burst 1509b/8 cburst 1739b/8 overhead 14b level 0
class htb 1:40 parent 1:1 leaf 400: prio 3 quantum 9000 rate 58000bit 
ceil 576000bit burst 1480b/8 cburst 1739b/8 overhead 14b level 0
class htb 1:50 parent 1:1 leaf 500: prio 7 quantum 2000 rate 20000bit 
ceil 576000bit burst 1461b/8 cburst 1739b/8 overhead 14b level 0
class htb 1:60 parent 1:1 leaf 600: prio 7 quantum 3000 rate 23000bit 
ceil 576000bit burst 1462b/8 cburst 1739b/8 overhead 14b level 0
class htb 1:70 parent 1:1 leaf 700: prio 7 quantum 1000 rate 14000bit 
ceil 576000bit burst 1458b/8 cburst 1739b/8 overhead 14b level 0
********************************************************

Now the classes 1:50 - 1:70 are often get much backlog, but the 
root-class 1:1 don''t reach the ceil rate. 
statistic looks like:
********************************************************
tc -s -d class show dev ppp0
class htb 1:1 root rate 576000bit ceil 576000bit burst 30Kb/8 mpu 0b 
overhead 0b cburst 1739b/8 mpu 0b overhead 14b level 7
 Sent 1485575598 bytes 3140554 pkts (dropped 0, overlimits 0)
 rate 480008bit 115pps
 lended: 1904616 borrowed: 0 giants: 0
 tokens: 385702 ctokens: -26458

class htb 1:10 parent 1:1 leaf 100: prio 0 quantum 7500 rate 58000bit 
ceil 115000bit burst 1480b/8 mpu 0b overhead 0b cburst 1508b/8 mpu 0b 
overhead 14b level 0
 Sent 1186471 bytes 15097 pkts (dropped 0, overlimits 0)
 rate 152bit
 lended: 15097 borrowed: 0 giants: 0
 tokens: 194207 ctokens: 99943

class htb 1:20 parent 1:1 leaf 200: prio 1 quantum 256 rate 282000bit 
ceil 576000bit burst 396b/2 mpu 0b overhead 0b cburst 543b/2 mpu 0b 
overhead 14b level 0
 Sent 39131574 bytes 884694 pkts (dropped 0, overlimits 0)
 rate 13296bit 39pps
 lended: 884643 borrowed: 51 giants: 0
 tokens: 8453 ctokens: 6229

class htb 1:30 parent 1:1 leaf 300: prio 2 quantum 9000 rate 117000bit 
ceil 576000bit burst 1509b/8 mpu 0b overhead 0b cburst 1739b/8 mpu 0b 
overhead 14b level 0
 Sent 1027775 bytes 5392 pkts (dropped 0, overlimits 0)
 rate 112bit
 lended: 5332 borrowed: 60 giants: 0
 tokens: 61194 ctokens: 15701

class htb 1:40 parent 1:1 leaf 400: prio 3 quantum 9000 rate 58000bit 
ceil 576000bit burst 1480b/8 mpu 0b overhead 0b cburst 1739b/8 mpu 0b 
overhead 14b level 0
 Sent 370952 bytes 750 pkts (dropped 0, overlimits 0)
 lended: 617 borrowed: 133 giants: 0
 tokens: 172179 ctokens: 21731

class htb 1:50 parent 1:1 leaf 500: prio 7 quantum 2000 rate 20000bit 
ceil 576000bit burst 1461b/8 mpu 0b overhead 0b cburst 1739b/8 mpu 0b 
overhead 14b level 0
 Sent 249243996 bytes 608136 pkts (dropped 0, overlimits 0)
 rate 88512bit 22pps
 lended: 133117 borrowed: 475019 giants: 0
 tokens: -439382 ctokens: 5148

class htb 1:60 parent 1:1 leaf 600: prio 7 quantum 3000 rate 23000bit 
ceil 576000bit burst 1462b/8 mpu 0b overhead 0b cburst 1739b/8 mpu 0b 
overhead 14b level 0
 Sent 831028684 bytes 1288890 pkts (dropped 62, overlimits 0)
 rate 278224bit 42pps backlog 38p
 lended: 154838 borrowed: 1134014 giants: 0
 tokens: -65884 ctokens: -21987

class htb 1:70 parent 1:1 leaf 700: prio 7 quantum 1000 rate 14000bit 
ceil 576000bit burst 1458b/8 mpu 0b overhead 0b cburst 1739b/8 mpu 0b 
overhead 14b level 0
 Sent 363652940 bytes 337633 pkts (dropped 0, overlimits 0)
 rate 100144bit 11pps
 lended: 42294 borrowed: 295339 giants: 0
 tokens: -421519 ctokens: 2886
********************************************************

Why the ceil rate can''t be reached? 
rate 480008bit from 576000bit a little bit to huge difference. And 
besides this i''m using the overhead patch from Jesper Dangaard Brouer 
(iproute+htb) which takes the ATM+AAL5+SSCS Overhead into account.
Can a slightly inaccurate clock has something todo with this?

Another question: why "tc show class" prints the overhead and mpu
value
twice? And why is the first overhead value = 0? 

-- 
Markus Schulz

Am Freitag, 10. Februar 2006 14:45 schrieb Markus Schulz:
> tc -s -d class show dev ppp0
> class htb 1:1 root rate 576000bit ceil 576000bit burst 30Kb/8 mpu 0b
> overhead 0b cburst 1739b/8 mpu 0b overhead 14b level 7
>  Sent 1485575598 bytes 3140554 pkts (dropped 0, overlimits 0)
>  rate 480008bit 115pps
>  lended: 1904616 borrowed: 0 giants: 0
>  tokens: 385702 ctokens: -26458
ok, i''ve understand now. Differenz comes from gross versus net data 
rates due to overhead of ATM-SAR and pppoe-overhead. All statistic 
values are netto values.

-- 
Markus Schulz

Markus Schulz wrote:
> Am Freitag, 10. Februar 2006 14:45 schrieb Markus Schulz:
> 
> 
>>tc -s -d class show dev ppp0
>>class htb 1:1 root rate 576000bit ceil 576000bit burst 30Kb/8 mpu 0b
>>overhead 0b cburst 1739b/8 mpu 0b overhead 14b level 7
>> Sent 1485575598 bytes 3140554 pkts (dropped 0, overlimits 0)
>> rate 480008bit 115pps
>> lended: 1904616 borrowed: 0 giants: 0
>> tokens: 385702 ctokens: -26458
> 
> 
> ok, i''ve understand now. Differenz comes from gross versus net
data
> rates due to overhead of ATM-SAR and pppoe-overhead. All statistic 
> values are netto values.
> 
Yes - you can get patches to do egress overheads for dsl - if you know 
your exact type and are prepared to rebuild kernel or the modules and tc.

Andy.

>> ok, i''ve understand now. Differenz comes from gross versus net
data
>> rates due to overhead of ATM-SAR and pppoe-overhead. All statistic 
>> values are netto values.
>>
> 
> Yes - you can get patches to do egress overheads for dsl - if you know 
> your exact type and are prepared to rebuild kernel or the modules and tc.
Oops didn''t see you were already patched - htb counters still see
things
at ip level.

FWIW You still can''t really use 576000 - one of my modems would do
it''s
aaal5/0 qos in whole cells so being slightly less, also if you 
start/restart the shaper on a live link the queue formed may never empty.

Running 286kbit on a 288kbit ATM rate line has never gone over for me 
and slowly drains if the modem buffer is full when started.

Andy.