Sinks for rtsp streaming quit error (CUDA failure:status=700)

• Hardware Platform GTX 1070
• DeepStream Version 5.0
• TensorRT Version 7.0
• NVIDIA GPU Driver Version 440

Description

I want to use deepsream-app run 2 config files to create 2 instances. And each instance has its own rtsp sink. My config is :

First config file

[sink0]
enable=1
type=4
codec=1
sync=0
bitrate=2000000
rtsp-port=8554
udp-port=5400

Second config file

[sink0]
enable=1
type=4
codec=1
sync=0
bitrate=2000000
rtsp-port=8555
udp-port=5401

But when I quit deepstream-app, it returns errors about cuda illegal memory access. When I cancle rtsp streaming sink group and set other sink type, it quits successfully.

Error info

...
** Deepstream: Quitting...**
Cuda failure: status=700
nvbufsurface: Error(-1) in releasing cuda memory
gstnvtracker: Got -1 destroying nvbufsurface 0x7fff3174c590
Cuda failure: status=700
nvbufsurface: Error(-1) in releasing cuda memory
gstnvtracker: Got -1 destroying nvbufsurface 0x7fff3174ae60
Cuda failure: status=700
nvbufsurface: Error(-1) in releasing cuda memory
gstnvtracker: Got -1 destroying nvbufsurface 0x7fff31749190
Cuda failure: status=700
nvbufsurface: Error(-1) in releasing cuda memory
gstnvtracker: Got -1 destroying nvbufsurface 0x7fff31746f80
ERROR: ../nvdsinfer/nvdsinfer_func_utils.cpp:31 [TRT]: engine.cpp (182) - Cuda Error in ~ExecutionContext: 700 (an illegal memory access was encountered)
ERROR: ../nvdsinfer/nvdsinfer_func_utils.cpp:31 [TRT]: INTERNAL_ERROR: std::exception
ERROR: ../nvdsinfer/nvdsinfer_func_utils.cpp:31 [TRT]: Parameter check failed at: ../rtSafe/safeContext.cpp::terminateCommonContext::165, condition: cudaEventDestroy(context.start) failure.
ERROR: ../nvdsinfer/nvdsinfer_func_utils.cpp:31 [TRT]: Parameter check failed at: ../rtSafe/safeContext.cpp::terminateCommonContext::170, condition: cudaEventDestroy(context.stop) failure.
ERROR: ../nvdsinfer/nvdsinfer_func_utils.cpp:31 [TRT]: ../rtSafe/safeRuntime.cpp (32) - Cuda Error in free: 700 (an illegal memory access was encountered)
terminate called after throwing an instance of 'nvinfer1::CudaError'
  what():  std::exception
[1] + Done                       "/usr/bin/gdb" --interpreter=mi --tty=${DbgTerm} 0<"/tmp/Microsoft-MIEngine-In-ihubn70r.x1k" 1>"/tmp/Microsoft-MIEngine-Out-bjq7jhld.9qf"

There is no update from you for a period, assuming this is not an issue any more.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

I can not reproduce the problem. Can you try the same case with T4 or Jetson platform?