[altq 20] RE: When a class is delayed (cbqstat)

[Fulvio Risso <F.Risso@cs.ucl.ac.uk>]

Sure.
But I don't think it's going to help you too much.
That's a trace where:
  CBQ: 2Mbps (ATM PVC)
  CLass: 25% and 75% (default); borrow not allowed
  MTU 9180
  Packet size: 9180
  Traffic: only one UDP flow in the default class (75%)
  input PVC: 3Mbps

Bye

	fulvio

================================================================
Class_ROOT for Interface pvc0:
        priority: 0 depth: 1 offtime: 1 [us] wrr_allot: 9259 bytes
        nsPerByte: 4000 (2.00 Mbps),    Measured: 1.32 [Mbps]
        pkts: 3586      bytes: 32953364
        overs: 0        overactions: 0
        borrows: 0      delays: 0
        drops: 0        drop_bytes: 0
        QCount: 0       (qmax: 30)
        AvgIdle: 1004 [us],     (maxidle: 1140 minidle: -1148 [us])

Class_DEFAULT for Interface pvc0:
        priority: 1 depth: 0 offtime: 42055 [us] wrr_allot: 14039 bytes
        nsPerByte: 5479 (1.46 Mbps),    Measured: 1.32 [Mbps]
        pkts: 3586      bytes: 32953364
        overs: 717      overactions: 717
        borrows: 0      delays: 717
        drops: 4234     drop_bytes: 38918928
        QCount: 30      (qmax: 30)
        AvgIdle: -19 [us],      (maxidle: 1819 minidle: -1572 [us])

Class_CONTROL for Interface pvc0:
        priority: 6 depth: 0 offtime: 5573437 [us] wrr_allot: 0 bytes
        nsPerByte: 200000       (40.00 Kbps),   Measured: 0.00 [Kbps]
        pkts: 0         bytes: 0
        overs: 0        overactions: 0
        borrows: 0      delays: 0
        drops: 0        drop_bytes: 0
        QCount: 0       (qmax: 30)
        AvgIdle: 242968 [us],   (maxidle: 242968 minidle: -57375 [us])

Class_0 on Interface pvc0:
        priority: 1 depth: 0 offtime: 341250 [us] wrr_allot: 4807 bytes
        nsPerByte: 16000        (500.00 Kbps),  Measured: 0.00 [Kbps]
        pkts: 0         bytes: 0
        overs: 0        overactions: 0
        borrows: 0      delays: 0
        drops: 0        drop_bytes: 0
        QCount: 0       (qmax: 30)
        AvgIdle: 14875 [us],    (maxidle: 14875 minidle: -4590 [us])


Class_ROOT for Interface pvc0:
        priority: 0 depth: 1 offtime: 1 [us] wrr_allot: 9259 bytes
        nsPerByte: 4000 (2.00 Mbps),    Measured: 1.32 [Mbps]
        pkts: 3676      bytes: 33780644
        overs: 0        overactions: 0
        borrows: 0      delays: 0
        drops: 0        drop_bytes: 0
        QCount: 0       (qmax: 30)
        AvgIdle: 1004 [us],     (maxidle: 1140 minidle: -1148 [us])

Class_DEFAULT for Interface pvc0:
        priority: 1 depth: 0 offtime: 42055 [us] wrr_allot: 14039 bytes
        nsPerByte: 5479 (1.46 Mbps),    Measured: 1.32 [Mbps]
        pkts: 3676      bytes: 33780644
        overs: 735      overactions: 734
        borrows: 0      delays: 734
        drops: 4340     drop_bytes: 39893280
        QCount: 29      (qmax: 30)
        AvgIdle: -19 [us],      (maxidle: 1819 minidle: -1572 [us])

Class_CONTROL for Interface pvc0:
        priority: 6 depth: 0 offtime: 5573437 [us] wrr_allot: 0 bytes
        nsPerByte: 200000       (40.00 Kbps),   Measured: 0.00 [Kbps]
        pkts: 0         bytes: 0
        overs: 0        overactions: 0
        borrows: 0      delays: 0
        drops: 0        drop_bytes: 0
        QCount: 0       (qmax: 30)
        AvgIdle: 242968 [us],   (maxidle: 242968 minidle: -57375 [us])

Class_0 on Interface pvc0:
        priority: 1 depth: 0 offtime: 341250 [us] wrr_allot: 4807 bytes
        nsPerByte: 16000        (500.00 Kbps),  Measured: 0.00 [Kbps]
        pkts: 0         bytes: 0
        overs: 0        overactions: 0
        borrows: 0      delays: 0
        drops: 0        drop_bytes: 0
        QCount: 0       (qmax: 30)
        AvgIdle: 14875 [us],    (maxidle: 14875 minidle: -4590 [us])



> -----Original Message-----
> From: owner-altq@csl.sony.co.jp [mailto:owner-altq@csl.sony.co.jp]On
> Behalf Of Kenjiro Cho
> Sent: Friday, June 04, 1999 6:08 PM
> To: Fulvio Risso
> Cc: altq@csl.sony.co.jp
> Subject: [altq 18] Re: When a class is delayed
> 
> 
> 
> Fulvio,
> 
> Sorry, I wasn't clear.
> Could you take the cbqstat output in the middle of your test scenario
> (when the class is sending faster than expected)?
> 
> -Kenjiro