After running deepstream for a period of time, the perf module stopped printing

Accelerated Computing Intelligent Video Analytics Metropolis Microservices for Jetson
,
1

Please provide complete information as applicable to your setup.

Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU) Jetson AGX
• DeepStream Version 7.0
**• JetPack Version (valid for Jetson only) 6.0 GA
**• TensorRT Version 8.6
**• NVIDIA GPU Driver Version (valid for GPU only) 540.3.0
**• Issue Type( questions, new requirements, bugs) questions
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)

After running deepstream for a period of time, the perf module stopped printing

2

This is the new topic from: Disconnecting after reaching the reconnection count, there is still information about this stream in printing - #25 by kesong

3

I have found a reproduction phenomenon,
Operation steps

  1. Wait for automatic deletion after adding the stream
  2. Add the stream again within 10 seconds after automatic deletion.
    And I saw this phenomenon in the log
    The screenshot below
    log20241101.zip (1.7 MB)
    批注 2024-11-01 102908
    批注 2024-11-01 1029081711×186 26.5 KB

0:05:25.695709822 e[32m 19e[00m 0xfffec00060c0 e[33;01mWARN e[00m e[00m nvmultiurisrcbin gstdsnvmultiurisrcbin.cpp:1509:s_stream_api_impl:<src_nvmultiurisrcbin>e[00m Failed to add sensor id=[29]; Already added

5

Excuse me, could you please help confirm if there are any issues with the tracking section of the configuration file
Many Thanks
config.txt (5.6 KB)

6

The config.txt seems fine. Maybe the issue related with add/remove camera. Is it possible to write one script to send add/remove camera command to reproduce the issue?

7

I have a question to consult,
Log print
Added stream : sub-batch = 1 seq index = 0 pad index = 2
Is adding a video stream to the tracking module a successful indicator ？
what do sub-batch seq index and

批注 2024-11-04 103323
批注 2024-11-04 103323761×454 52 KB

pad index mean?

8

Because I found that the number of printed
Added streams
is greater than the actual number of added streams

9

The log come from: /opt/nvidia/deepstream/deepstream-7.0/sources/gst-plugins/gst-nvtracker/nvtracker_proc.cpp.
sub-batch is the index of sub-batch in nvtracker configure file. seq index means the number index in the sub-batch. pad index means the sink pad index of nvstreammux.

10

Is it a successful indicator about adding a video stream to the tracking module？

11

Yes, I suspect the issue is related with the sleep time between add/remove stream. Can you use one script to add/remove stream? So you can modify the sleep time to verify the issue.

12

Yes I try it

13

Found the steps for 100% reproduction

  1. Ensure that the maximum quantity in the track is 8
  2. Write a script that uses web APIs to add 7 streams
  3. Execute script
  4. Select a flow to delete
  5. Wait for a period of time to continue executing 3 and 4 (combined execution), and after executing 2-4 times, this phenomenon occurs
14

It has already appeared 2-3 times

15

Can you share the test script which can reproduce the issue.

16

test.txt (2.2 KB)

17

Excuting 3 will add 7 streams, but excuting 4 only remove one stream. How to ensure the max stream is 8?

18

Are you referring to mutiuri_strc-bin or tracker?
Repeat execution 3 （Repeatedly using this script afterwards）

Because the configfile is only configed with 8 streams

19

I suspect that when Deepstream deleted the stream, Track did not delete the video stream

20

Yes, not sure if you can reproduce the issue with below steps:

  1. add 8 streams
  2. remove one stream
  3. sleep 30 seconds
  4. add one stream

You can repeat 2-3-4 steps.

22

I tried three times yesterday and this phenomenon all appeared.
If it weren’t for the deepstreamtest5 issue, I don’t know if it was caused by code changes here