Pipeline restarts without any error message

Accelerated Computing Intelligent Video Analytics DeepStream SDK
,
1

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU): NVIDIA Jetson Xavier NX - AVerMedia NX215
• DeepStream Version: 6.2
• JetPack Version (valid for Jetson only): 5.1
• TensorRT Version: 8.5.2.2
• Issue Type: Bug

image
image1067×2027 131 KB

The Image above is the structure for my currently deepstream pipeline. Recently I made some changes to my pipeline to integrate MediaMTX for webrtc. But for that i go issues regarding DTS/PTS so I integrated an ffmpeg relay in between to resolve that issue and made changes to the encoder.

 # Encoder for RTSP Branch

            encoder_rtsp = self.make_element("nvv4l2h264enc", "encoder-rtsp", i)

            encoder_rtsp.set_property("bitrate", 4000000)

            # WebRTC compatibility

            encoder_rtsp.set_property("profile", 0)         # Baseline

            # encoder_rtsp.set_property("num-B-Frames", 0)    # No B-frames

            encoder_rtsp.set_property("poc-type", 2)         # Force decode order = display order (prevents non-monotonic PTS/DTS)


            # Stream stability

            encoder_rtsp.set_property("iframeinterval", 30) # GOP

            encoder_rtsp.set_property("idrinterval", 30)    # Force IDR every GOP (important)

            encoder_rtsp.set_property("insert-sps-pps", 1)  # Required for many players

            encoder_rtsp.set_property("insert-aud", 1)      # Often helps WebRTC/browser decoders


            # Performance

            encoder_rtsp.set_property("preset-level", 1)    # UltraFast

            encoder_rtsp.set_property("maxperf-enable", 1)  # Max performance mode

            self.pipeline.add(encoder_rtsp)

In my pipeline I save videos 15 mins each, and the pipeline runs for 12 hrs per day. Since I have pushed the change at start everything works fine, but after 8-10 mins the whole pipeline restart without throwing any error. The pipeline is set to restart automatically with the FPS drop to 0 for more than 25 secs. So that event occurs and everything is restarted and videos that are created, get corrupted too.

3

Can you elaborate what kind of restart it is?

Is there any detailed log or other information? There is no clue in your description from this post.

4

The error was due to my .engine file which I used for detection. But the pipeline restarted only on devices which had a different .engine file.

But it would be great if you could give me ways, I get the error logged in. As I did not get any error, so I started reverting back my changes until the pipeline issue was resolved.

5

What do you mena by “restart”? If the .engine file caused the problem, what is the difference of the .engine file compared to the original .engine file you used?