• Hardware Platform (Jetson / GPU) Jetson AGX Orin • DeepStream Version 6.2/7 • JetPack Version (valid for Jetson only) 6.0
Hello,
I’m running an nvmultiurisrcbin pipeline with smart recording in Python. I have it working correctly with additional bindings for sessionID, and I stop my recordings by emitting the stop-sr signal with 0, as shown in an example I found on the forum.
However, I’m encountering an issue where my 4-camera pipeline exits with a segmentation fault error (exit code 139) after recording for a variable duration, ranging from 1 hour to a few days.
I would like to understand why this might be happening. Could it be related to smart recording not working correctly in Python, or is there a potential issue with how I emit the signals? There is also a possibility that I emit the start-sr signal 20 ms after emitting the stop-sr signal. Could this timing be the reason for the segmentation fault?
Any help would be greatly appreciated!
EDIT: It is nvurisrcbin pipeline, not nvmultiurisrcbin pipeline, sorry for confusion.
please refer to this topic for how to use smart recording of nvmultiurisrcbin.
2.which sample are you testing or referring to? To narrow this issue, if using two sources, will the app crash with the same error?
I apologize for the confusion. I was actually experiencing the same issues with the nvurisrcbin pipeline. In fact, I intended to ask about the nvurisrcbin pipeline but mistakenly mentioned nvmultiurisrcbin. While I managed to get smart recording to work with nvmultiurisrcbin using your provided example, the same issues are still present.
For simplicity, I’ve created a separate pipeline specifically for recording videos. It includes four nvurisrcbin elements connected to fakesinks. I trigger the sr-start and sr-stop signals from my main pipeline (i send a message over the network to know then to start and stop recording), and while it works as expected most of the time, occasional crashes still occur. The frequency of these crashes varies depending on the device; on some devices, it almost never happens. This leads me to believe it might be related to camera stability or how often I emit signals.
The issue is challenging to troubleshoot as segmentation faults do not provide any meaningful feedback. One potential cause I’m considering is the timing of the signals—could sending the sr-start signal before receiving the sr-done callback be causing this problem?
sr-done will be received immediately after triggering sr-stop. you can trigger sr-start after receiving sr-done.
please simplify the application to narrow down this issue. here are some suggestions.
if using two nvurisrcbin, will the app crash?
if using four nvurisrcbin without recording, will the app crash?
There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks