Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

nffPktgen sends traffic with no consistent speed and pkts count #694

Open
Bubbachuck opened this issue Feb 5, 2020 · 7 comments
Open

nffPktgen sends traffic with no consistent speed and pkts count #694

Bubbachuck opened this issue Feb 5, 2020 · 7 comments

Comments

@Bubbachuck
Copy link

Hi friends,
When I use nffPktgen sending traffic, I found the speed is not consistent and the pkts count is also not consistent.
Could you please share me how I can send traffic with a consistent speed?
And sometime below information is observed:
WARNING: Can't start new clone for segment1 instance 0

What is the reason for this warning?

Thanks.

@aregm
Copy link
Owner

aregm commented Feb 5, 2020

Hi @Bubbachuck - one of the reasons for such behavior can be an insufficient number of cores - how many cores do you have and what is the environment?

@Bubbachuck
Copy link
Author

Thanks @aregm for your response.
I run it in a VM, which has 16 cores in a NUMA, corresponding to 16 HTs in host server.

EAL: Detected 16 lcore(s)
EAL: Detected 1 NUMA nodes
EAL: Multi-process socket /var/run/dpdk/rte/mp_socket
EAL: Selected IOVA mode 'PA'

@aregm
Copy link
Owner

aregm commented Feb 6, 2020

@ifilippov

@aregm
Copy link
Owner

aregm commented Feb 6, 2020

@Bubbachuck can you please specify the environment and the command you are running nffpktgen?

@ifilippov
Copy link
Contributor

Hi @Bubbachuck! Could you also please describe what speed do you want to achieve:
Is pktgen able to achieve it before began to be unstable?
When pktgen is unstable does it make speed greater or lesser than expected?

Scheduler is designed in the way that it supposes that tasks at different cores will not compete with each other - is this satisfied in your VM environment? If not - you can try to make speed lower, than scheduler will not use additional cores.

@Bubbachuck
Copy link
Author

@aregm The command is "trafficgen" which is in the folder of "gtp-u". The environment is a VM with 16 x 1G hugepages, and 16 cores in a NUMA from 16HTs in a socket.

@Bubbachuck
Copy link
Author

Hi @ifilippov The target speed is 1500000 PKT/S. It looks can achieve the speed before to be unstable. I am not sure the real-time speed, but packets drop is found in debug message.
In additional, how can I count the packets sent out?

Please find below debug information:

DEBUG: System is using 16 cores now. 0 cores are left available.
DEBUG: Current speed of fast generator1 is 1687104 PKT/S, target speed is 1500000 PKT/S
DEBUG: Current speed of 0 instance of segment1 is 1668222 PKT/S, cloneNumber: 2 queue number: 1
DEBUG: Current speed of 0 instance of segment2 is 1045878 PKT/S, cloneNumber: 1 queue number: 16
DEBUG: Current speed of fast generator2 is 1192384 PKT/S, target speed is 1500000 PKT/S
DEBUG: Current speed of 0 instance of segment3 is 1086526 PKT/S, cloneNumber: 1 queue number: 1
DEBUG: Current speed of 0 instance of segment4 is 0 PKT/S, cloneNumber: 1 queue number: 16
DROP: Flow functions together dropped 366553 packets
WARNING: Can't start new clone for segment1 instance 0

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants