Re: [tcpPrague] TSO burst sizing causing TCP Prague unfairness on high capacity links ?

[Ashutosh Srivastava <[e-mail]>]

Hi Olivier,

Please find comments inline.

Could you confirm that this is also happening with the tip of the current
> Branch? I.e., 3b63cc0
>

Yes, I observe these unfairness issues with the tip of the current branch (
3b63cc0). This happens both in the case of a flow entering after some gap
or with multiple flows starting together.

- Confirm that this happens with both the internal TCP pacing and the fq
> qdisc as pacer on the data sender
>
Yes , it happens with both  with internal TCP pacing and fq pacing.

- Confirm that this happens both with and without gro/gso/tso/lro on the
> endhosts (data sender and receiver, and also the aqm node as fq does not do
> gro splitting prior to enqueue)?
>

This *does not happen* when I turn off gro/gso/tso on all the nodes (end
hosts, AQM) in my experiment.

- Confirm that the problem persists if you increase the base RTT--a simple
> netem qdisc on the reverse path to add 2-5ms should be sufficient, do not
> forget to disable gro/gso as that poorly interacts with netem.
>

The problem persists even with some added delay.


> - Log the reported CE marks by the AQM, as well as the
> delivered_ce/received_ce counters throughout the experiment on the data
> sender/receiver?
>

I have uploaded  additional data for experiments with commit 3b63cc0 on the
google drive I shared earlier :
https://drive.google.com/drive/folders/1pLC0dcMF0-M1cgtw9IhoFiYOMvFJ-cc7?usp=sharing

The columns of the AQM csv file are : timestamp, dropped packets, queue
backlog (bytes), queue backlog (packets) , CE_marks
Columns of the ss data file at the sender are : timestamp, ID, cwnd, SRTT
(ms), delivered, delivered_ce, pacing rate (Mbps)

I have also added the raw data files. The experiment settings were the same
as in my last mail.

Thank you,
Ashutosh.