Our application was originally built on Jetpack 4.6 using nvidia-tensorrt v8.0.1-1+cuda10.2, but our latest unit in the field uses Jetpack 4.6.3 with nvidia-tensorrt v8.2.1-1+cuda10.2 and I’m running into the below errors with gstreamer (GST) while running our application there.
(python3:11146): GStreamer-CRITICAL **: 07:36:32.052: gst_mini_object_unlock: assertion 'state >= SHARE_ONE' failed
(python3:11146): GStreamer-CRITICAL **: 07:36:32.053: gst_mini_object_unref: assertion 'GST_MINI_OBJECT_REFCOUNT_VALUE (mini_object) > 0' failed
(python3:11146): GStreamer-CRITICAL **: 07:36:32.121: gst_mini_object_unlock: assertion 'state >= SHARE_ONE' failed
(python3:11146): GStreamer-CRITICAL **: 07:36:32.170: gst_mini_object_unlock: assertion 'state >= SHARE_ONE' failed
(python3:11146): GStreamer-CRITICAL **: 07:36:32.170: gst_mini_object_unref: assertion 'GST_MINI_OBJECT_REFCOUNT_VALUE (mini_object) > 0' failed
(python3:11146): GStreamer-CRITICAL **: 07:36:32.204: gst_mini_object_unlock: assertion 'state >= SHARE_ONE' failed
(python3:11146): GStreamer-CRITICAL **: 07:36:32.204: gst_mini_object_unref: assertion 'GST_MINI_OBJECT_REFCOUNT_VALUE (mini_object) > 0' failed
(python3:11146): GStreamer-CRITICAL **: 07:36:32.256: gst_mini_object_unlock: assertion 'state >= SHARE_ONE' failed
(python3:11146): GStreamer-CRITICAL **: 07:36:32.259: gst_mini_object_unref: assertion 'GST_MINI_OBJECT_REFCOUNT_VALUE (mini_object) > 0' failed
(python3:11146): GStreamer-CRITICAL **: 07:36:32.295: gst_mini_object_unlock: assertion 'state >= SHARE_ONE' failed
After the above, it exits with a Segmentation Fault
Can I install nvidia-tensorrt v8.0.1-1+cuda10.2 on jetpack 4.6.3?
or
Are the above errors pointing to something else being the problem?
We need to fix this ASAP
EDIT: We have another system that’s working on v8.2.1-1+cuda10.2. I don’t know where these errors are coming from. I’ve tried fully reinstalling gstreamer, rebuilding our model on the local system, and reinstalling the deepstream prereq’s with no luck.