pfc.txt

[qizhouzhang]

Hi yuliang. I have try three very large flow have same dst, but the pfc.txt is empty? cc is DCQCN? Is it normal?

[liyuliang001]

How about try more flows, say 40?

On Mon, Dec 7, 2020 at 5:39 AM qizhouzhang [email protected] wrote:

Hi yuliang. I have try three very large flow have same dst, but the pfc.txt is empty? cc is DCQCN? Is it normal?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/alibaba-edu/High-Precision-Congestion-Control/issues/14, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIWJVCHNH7MJCL6BGV7QSTSTS5I5ANCNFSM4UQKWXJQ .

[qizhouzhang]

when I try flow number is 100, the program was terminated with this error "Command ['/home/gdp/hpcc/simulation/build/scratch/third', 'mix/config.txt'] terminated with signal SIGSEGV", Have you met before？ And when flow is number is 50, the pfc.txt is still empty

[qizhouzhang]

I understand why SIGSEGV. I have try 255 flow-> 1 dst。PFC.txt is empty。。。

[liyuliang001]

How about try analysis/trace_reader? That will tell you all packet-level events, including the queue length when each packet arrive at each hop. So you can see what happened.

On Mon, Dec 7, 2020 at 8:50 PM qizhouzhang [email protected] wrote:

I understand why SIGSEGV. I have try 255 flow-> 1 dst。PFC.txt is empty。。。

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/alibaba-edu/High-Precision-Congestion-Control/issues/14#issuecomment-740325904, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIWJVBO4CC4SOHNUMX2QVLSTWH65ANCNFSM4UQKWXJQ .

[qizhouzhang]

Let me see see. Thanks!

[qizhouzhang]

I have tried the (DCQCN + enable win), the PFC is not triggered. But only with DCQCN, the PFC is triggered. Is it reasonable?👀

[liyuliang001]

Yes. This is expected :) You can refer to HPCC paper, and you will find the same conclusion.

On Tue, Dec 8, 2020 at 4:06 AM qizhouzhang [email protected] wrote:

I have tried the (DCQCN + enable win), the PFC is not triggered. But only with DCQCN, the PFC is triggered. Is it reasonable?👀

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/alibaba-edu/High-Precision-Congestion-Control/issues/14#issuecomment-740521623, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIWJVBZADYV3IVGV3NLTULSTX3EFANCNFSM4UQKWXJQ .

[qizhouzhang]

thanks

[qizhouzhang]

Hi ! I have another question. In traffic_gen.py has param t and in config.txt also have the simulation stop time. The param t should equal the simu stop time in config.txt?

[liyuliang001]

Hi, These 2 parameters are independent. If simulation stop time is earlier than t, simulations will stop without running all flows (since some flows are indicated to start after simulation stop). Normally in my setting, t is smaller than simulation stop time, to guarantee all traffic that I want is generated.

On Thu, Dec 10, 2020 at 1:29 AM qizhouzhang [email protected] wrote:

Reopened #14 https://github.com/alibaba-edu/High-Precision-Congestion-Control/issues/14 .

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/alibaba-edu/High-Precision-Congestion-Control/issues/14#event-4093582961, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIWJVFE2YRO4J7LQW5SBVLSUB2FJANCNFSM4UQKWXJQ .

[qizhouzhang]

I run a simulation about 10W flows. It take about 5 h. If I use traffic_gen.py generate flows with 320hosts at load 0.7, flow are 1600W. It seems to need 5 * 160 = 800h / 24 = 34d. Is there any way to speed up this process?😂

[liyuliang001]

Yes, simulation is slow :) In my experience, running 320 hosts, at load 0.3 (or 0.5?), it takes 1 day to simulate 0.1s~0.2s.

But your number seems too slow to me. What's the t in your setting?

On Thu, Dec 10, 2020 at 9:55 AM qizhouzhang [email protected] wrote:

I run a simulation about 10W flows. It take about 5 h. If I use traffic_gen.py generate flows with 320hosts at load 0.7, flow are 1600W. It seems to need 5 * 160 = 800h / 24 = 34d. Is there any way to speed up this process?😂

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/alibaba-edu/High-Precision-Congestion-Control/issues/14#issuecomment-742609310, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIWJVD6FSML3XBY3RR4HPTSUDVPBANCNFSM4UQKWXJQ .

[qizhouzhang]

I set t to 5 second. I will try smaller t👌😂 thanak you.

[qizhouzhang]

Hi Yuliang! I want to know how to calculate the real-time throughput.Suppose a single flow's path is 1(host)->301(switch)->2(Host)?

[liyuliang001]

I think you can capture the packet trace, and then calculate the throughput yourself. That's what I did :)

To capture packet trace, you can check https://github.com/alibaba-edu/High-Precision-Congestion-Control/blob/master/simulation/run.py#L76 With this option set, the simulation also output a trace file to: https://github.com/alibaba-edu/High-Precision-Congestion-Control/blob/master/simulation/run.py#L13

Then you can use https://github.com/alibaba-edu/High-Precision-Congestion-Control/blob/master/analysis/trace_reader.cpp to read the trace file.

[songhexiang]

I have tried the (DCQCN + enable win), the PFC is not triggered. But only with DCQCN, the PFC is triggered. Is it reasonable?👀

So what is the test case when PFC is triggered, what is the scale of the test? I have tried websearch 30%load for 1 second and only with DCQCN, but the pfc.txt is empty.

[zqz-net]

You can try multi to one traffic pattern. (Multi source hosts to one destination host at the same time)

[SOULOFCINDERS]

when I try flow number is 100, the program was terminated with this error "Command ['/home/gdp/hpcc/simulation/build/scratch/third', 'mix/config.txt'] terminated with signal SIGSEGV", Have you met before？ And when flow is number is 50, the pfc.txt is still empty

Hello,I also meet the error "Command ['/home/gdp/hpcc/simulation/build/scratch/third', 'mix/config.txt'] terminated with signal SIGSEGV",have you got the reason of the error?

[0-polarstar]

我试过（DCQCN + enable win），PFC没有触发。但只有DCQCN才会触发 PFC。合理吗？👀

那么PFC触发时的测试用例是什么，测试的规模是多少？我已经尝试了 websearch 30%load 1 秒，并且只使用 DCQCN，但是 pfc.txt 是空的。

我也想问，试了好几组数据都无法触发pfc

[liyuliang001]

Yes, you can check the paper Figure 11.

On Sun, May 1, 2022 at 5:45 AM 0-polarstar @.***> wrote:

我试过（DCQCN + enable win），PFC没有触发。但只有DCQCN才会触发 PFC。合理吗？👀

那么PFC触发时的测试用例是什么，测试的规模是多少？我已经尝试了 websearch 30%load 1 秒，并且只使用 DCQCN，但是 pfc.txt 是空的。

我也想问，试了好几组数据都无法触发pfc

— Reply to this email directly, view it on GitHub https://github.com/alibaba-edu/High-Precision-Congestion-Control/issues/14#issuecomment-1114226718, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAIWJVDCQQR2F4TAVZZVAETVHZ4G5ANCNFSM4UQKWXJQ . You are receiving this because you commented.Message ID: @.*** .com>

[zqz-net]

我试过（DCQCN + enable win），PFC没有触发。但只有DCQCN才会触发 PFC。合理吗？👀

那么PFC触发时的测试用例是什么，测试的规模是多少？我已经尝试了 websearch 30%load 1 秒，并且只使用 DCQCN，但是 pfc.txt 是空的。

我也想问，试了好几组数据都无法触发pfc

You can tune the pfc threshold to trigger PFC in (switch_mmu.cc file)

[0-polarstar]

可以加您联系方式吗？我对阈值设置不是很确定

------------------ 原始邮件 ------------------ 发件人: "alibaba-edu/High-Precision-Congestion-Control" @.>; 发送时间: 2022年5月2日(星期一) 上午8:29 @.>; 抄送: "Last of the @.@.>; 主题: Re: [alibaba-edu/High-Precision-Congestion-Control] pfc.txt (#14)

我试过（DCQCN + enable win），PFC没有触发。但只有DCQCN才会触发 PFC。合理吗？👀

那么PFC触发时的测试用例是什么，测试的规模是多少？我已经尝试了 websearch 30%load 1 秒，并且只使用 DCQCN，但是 pfc.txt 是空的。

我也想问，试了好几组数据都无法触发pfc

You can tune the pfc threshold to trigger PFC in (switch_mmu.cc file)

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[zqz-net]

how about your email?