Deepstream reconnection has a chance to trigger blocking

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) GPU
• DeepStream Version5.1
• JetPack Version (valid for Jetson only)
• TensorRT Version 7.2
• NVIDIA GPU Driver Version (valid for GPU only) 440
• Issue Type( questions, new requirements, bugs) bug
• 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)
This is a blocking situation.

This is the code, gst_element_set_state (src_bin->bin,
GST_STATE_NULL) triggering a block.


gboolean
reset_source_pipeline (gpointer data)
{
  NvDsSrcBin *src_bin = (NvDsSrcBin *) data;
  GstState state, pending;
  GstStateChangeReturn ret;
    NVGSTDS_INFO_MSG_V ("1111111111111111111");
  g_mutex_lock(&src_bin->bin_lock);
    NVGSTDS_INFO_MSG_V ("1111111111111111111.5");
  gettimeofday (&src_bin->last_buffer_time, NULL);
  gettimeofday (&src_bin->last_reconnect_time, NULL);
    NVGSTDS_INFO_MSG_V ("1111111111111111111.6");
  g_mutex_unlock(&src_bin->bin_lock);
    NVGSTDS_INFO_MSG_V ("2222222222222222");
  if (gst_element_set_state (src_bin->bin,
          GST_STATE_NULL) == GST_STATE_CHANGE_FAILURE) {
      NVGSTDS_INFO_MSG_V ("22222222.5");
    GST_ERROR_OBJECT (src_bin->bin, "Can't set source bin to NULL");
    return FALSE;
  }
    NVGSTDS_INFO_MSG_V ("3333333333333333");
  NVGSTDS_INFO_MSG_V ("Resetting source %d", src_bin->bin_id);

  GST_CAT_INFO (NVDS_APP, "Reset source pipeline %s %p\n,", __func__, src_bin);
  if (!gst_element_sync_state_with_parent (src_bin->bin)) {
    GST_ERROR_OBJECT (src_bin->bin, "Couldn't sync state with parent");
  }

  ret = gst_element_get_state (src_bin->bin, &state, &pending, 0);

  GST_CAT_DEBUG (NVDS_APP, "Bin %d %p: state:%s pending:%s ret:%s",
          src_bin->bin_id, src_bin, gst_element_state_get_name(state),
          gst_element_state_get_name(pending),
          gst_element_state_change_return_get_name(ret));

  if (ret == GST_STATE_CHANGE_ASYNC || ret == GST_STATE_CHANGE_NO_PREROLL) {
    if (!src_bin->async_state_watch_running)
      g_timeout_add (20, watch_source_async_state_change, src_bin);
    src_bin->async_state_watch_running = TRUE;
    src_bin->reconfiguring = TRUE;
  } else if (ret == GST_STATE_CHANGE_SUCCESS && state == GST_STATE_PLAYING) {
    src_bin->reconfiguring = FALSE;
  }
  return FALSE;
}

Can you have a try with latest version DS 6.0?

I compared 6.0 to the reset_source_pipeline function without any changes

Can you share more about your use case?

I use deepstream-test5 through “./deepstream-test5-app -c configs/test5_config_file_src_infer.txt” command.
No other changes.

Can you run below command line before run your test application to get some log?

$ export GST_DEBUG=3


I have upgraded to 6.0.
This question may involve a change of mine. I tried to use unlink_element_from_streammux_sink_pad(bin->streammux, bin->sub_bins[i].bin) to delete the stream dynamically and blocked.And then the reconnection is blocked.

Can you refer below for dynamical add/remove source?

deepstream_reference_apps/deepstream_test_rt_src_add_del.c at master · NVIDIA-AI-IOT/deepstream_reference_apps (github.com)

Yes, I already did.
But there was another problem.
I added stop_release_source to deepstream_test5_app_main,c, it is work.
But it cannot be called in the bus_callback function, otherwise it would block.

• Hardware Platform (Jetson / GPU) GPU
• DeepStream Version 6.2
• JetPack Version (valid for Jetson only)
• TensorRT Version 8.5
• NVIDIA GPU Driver Version (valid for GPU only) 525

Anything new? I got this bug. Can you help me? @kesong @549981178

I am trying to solve this issue for several months now without any success. It happens in all DS versions.

Can you verify it with DS sample? deepstream_reference_apps/runtime_source_add_delete/deepstream_test_rt_src_add_del.c at master · NVIDIA-AI-IOT/deepstream_reference_apps (github.com)

I tried it with DS sample but not work when increase time in g_timeout_add_seconds from 15 to 30.
It happens when use rtspsrc in deepstream_source_bin.c to connect rtsp.
Can you focus reset_source_pipeline to fix it? @kesong

Can you reproduce the issue on latest DS version 6.3 without any changes?

I tried with latest DS version 6.3 without any changes. It happens in all DS versions. @kesong

Anything new? Can you help me? @kesong

Please submit one new topic as the original is report on Jan 05 2022. Thanks.

I meet this question too, but no solution, I find use rtsp source live is ok, but live555 rtsp server will block when 2 pipeline is running, only one pipeline also not block, you can try change source.