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

Message delays when network under load #5552

Open
1 task done
yuzu-ogura opened this issue Jan 9, 2025 · 1 comment
Open
1 task done

Message delays when network under load #5552

yuzu-ogura opened this issue Jan 9, 2025 · 1 comment
Labels
need more info Issue that requires more info from contributor

Comments

@yuzu-ogura
Copy link

yuzu-ogura commented Jan 9, 2025

Is there an already existing issue for this?

  • I have searched the existing issues

Expected behavior

as soon as possible to recovery sending message after occupation network.

Current behavior

During sending message, we maliciously occupied network for 10 seconds, after occupation network, it takes at least 30 seconds to recovery sending message.

Steps to reproduce

  1. start a subscriber
  2. start a publisher
  3. network bandwidth is nearly 10MB/s, message size is 2MB, each message sending will cost 300 ms.
  4. after communication successful, use iperf to occupation network, it will cost 10 seconds, the data send will block in the meanwhile
  5. after occupation network, it takes at least 30 seconds to recovery sending message.
  6. and then message send normally.

Fast DDS version/commit

Fast-RTPS version: 2.6.8

Platform/Architecture

Ubuntu Focal 20.04 amd64

Transport layer

UDPv4

Additional context

Im not sure whether my scene is similar with #1062 ?

XML configuration file

No response

Relevant log output

No response

Network traffic capture

No response

@yuzu-ogura yuzu-ogura added the triage Issue pending classification label Jan 9, 2025
@Javgilavi
Copy link
Contributor

Hi @yuzu-ogura,

Thank you for reporting this issue!

It seems that the problem might be related to an accumulation of messages during periods of high network usage, causing significant delays once the messages can be received again.

There is a data member in the UDP Transport Description that could address this issue. Please try using non_blocking_send.

When the bandwidth is saturated, this option ensures that a sent message is assumed to be lost instead of accumulating during the high-occupancy period.

If this does not resolve the issue, please share the QoS settings you are using in this scenario, and I will follow up to further investigate the problem.

@Javgilavi Javgilavi added need more info Issue that requires more info from contributor and removed triage Issue pending classification labels Jan 10, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
need more info Issue that requires more info from contributor
Projects
None yet
Development

No branches or pull requests

2 participants