Stream stops working after a few seconds

[breqdev]

Hi there!

I'm having an issue using rosimagesrc with the usb_cam package. Right now, my setup includes:

• An Arducam B0292 camera
• The usb_cam_node_exe node running, outputting /usb_cam/image_raw and /usb_cam/image_raw/compressed (among other topics)
• A GStreamer pipeline using rosimagesrc to pull images from usb_cam_node_exe.
• RMW_IMPLEMENTATION set to rmw_cyclonedds_cpp, and all nodes running on the same host

I'm able to run rqt_image_view and subscribe to /usb_cam/image_raw/compressed to view the camera stream continuously, and I'm also able to ros2 topic echo /usb_cam/image_raw to validate that image data is being published continuously on that topic. I get the expected framerate of around 30 fps.

However, when I try to run the example pipeline (or any pipeline with rosimagesrc) on this input, the pipeline works normally for a few seconds and then stalls:

Full command output

rover@base-station:~/rover/ros2_ws (main) $ gst-launch-1.0 --gst-plugin-path=/home/rover/rover/submodule_ws/install/gst_bridge/lib/gst_bridge -v rosimagesrc ros-topic="/usb_cam/image_
raw" ! gamma gamma=2.0 ! rosimagesink ros-topic="image_gamma_corrected"
Setting pipeline to PAUSED ...
[WARN] [1737742017.110227812] [rclcpp]: logging was initialized more than once
Pipeline is live and does not need PREROLL ...
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
[INFO] [1737742017.113722520] [gst_image_src_node]: getcaps with filter 'NULL'
[INFO] [1737742017.113760172] [gst_image_src_node]: waiting for first message
[INFO] [1737742017.113740118] [gst_image_src_node]: stream_start at 1737742017113739577
New clock: GstSystemClock
^Chandling interrupt.
Interrupt: Stopping pipeline ...
Execution ended after 0:00:22.872195157
Setting pipeline to NULL ...
^C
rover@base-station:~/rover/ros2_ws (main) $ gst-launch-1.0 --gst-plugin-path=/home/rover/rover/submodule_ws/install/gst_bridge/lib/gst_bridge -v rosimagesrc ros-topic="/usb_cam/image_raw" ! gamma gamma=2.0 ! rosimagesink ros-topic="image_gamma_corrected"
Setting pipeline to PAUSED ...
[WARN] [1737742041.009261417] [rclcpp]: logging was initialized more than once
Pipeline is live and does not need PREROLL ...
[INFO] [1737742041.013471393] [gst_image_src_node]: getcaps with filter 'NULL'
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
[INFO] [1737742041.013504321] [gst_image_src_node]: waiting for first message
[INFO] [1737742041.013539688] [gst_image_src_node]: stream_start at 1737742041013538869
New clock: GstSystemClock
[INFO] [1737742041.022241014] [gst_image_src_node]: preparing video with caps 'video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1'
[INFO] [1737742041.022349229] [gst_image_sink_node]: preparing video with caps 'video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1'
[INFO] [1737742041.022364850] [gst_image_sink_node]: setcaps format string is RGB 
[INFO] [1737742041.022370069] [gst_image_sink_node]: setcaps n_components is 3
[INFO] [1737742041.022374342] [gst_image_sink_node]: setcaps bits is 8
[INFO] [1737742041.022378514] [gst_image_sink_node]: setcaps pixel_stride is 3
/GstPipeline:pipeline0/Rosimagesrc:rosimagesrc0.GstPad:src: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
/GstPipeline:pipeline0/GstGamma:gamma0.GstPad:src: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
/GstPipeline:pipeline0/Rosimagesink:rosimagesink0.GstPad:sink: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
/GstPipeline:pipeline0/GstGamma:gamma0.GstPad:sink: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
Redistribute latency...
WARNING: from element /GstPipeline:pipeline0/Rosimagesink:rosimagesink0: Pipeline construction is invalid, please add queues.
Additional debug info:
../libs/gst/base/gstbasesink.c(1249): gst_base_sink_query_latency (): /GstPipeline:pipeline0/Rosimagesink:rosimagesink0:
Not enough buffering available for  the processing deadline of 0:00:00.020000000, add enough queues to buffer  0:00:00.020000000 additional data. Shortening processing latency to 0:00:00.000000000.
WARNING: from element /GstPipeline:pipeline0/Rosimagesink:rosimagesink0: Pipeline construction is invalid, please add queues.
Additional debug info:
../libs/gst/base/gstbasesink.c(1249): gst_base_sink_query_latency (): /GstPipeline:pipeline0/Rosimagesink:rosimagesink0:
Not enough buffering available for  the processing deadline of 0:00:00.020000000, add enough queues to buffer  0:00:00.020000000 additional data. Shortening processing latency to 0:00:00.000000000.
[INFO] [1737742041.023934638] [gst_image_sink_node]: stream_start at 1737742041023931735
0:00:01.5 / 99:99:99.

Since a few of the error messages point to queue elements being needed, I tried adding those to the pipeline. This removed those errors but the pipeline still stopped working after a few seconds:

rover@base-station:~/rover/ros2_ws (main) $ gst-launch-1.0 --gst-plugin-path=/home/rover/rover/submodule_ws/install/gst_bridge/lib/gst_bridge -v rosimagesrc ros-topic="/usb_cam/image_raw" ! queue ! gamma gamma=2.0 ! queue ! rosimagesink ros-topic="image_gamma_corrected"
Setting pipeline to PAUSED ...
[WARN] [1737742649.164740571] [rclcpp]: logging was initialized more than once
Pipeline is live and does not need PREROLL ...
[INFO] [1737742649.168812727] [gst_image_src_node]: getcaps with filter 'NULL'
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
[INFO] [1737742649.168850048] [gst_image_src_node]: waiting for first message
[INFO] [1737742649.168911356] [gst_image_src_node]: stream_start at 1737742649168910494
New clock: GstSystemClock
[INFO] [1737742649.173977409] [gst_image_src_node]: preparing video with caps 'video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1'
/GstPipeline:pipeline0/Rosimagesrc:rosimagesrc0.GstPad:src: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
/GstPipeline:pipeline0/GstQueue:queue0.GstPad:sink: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
/GstPipeline:pipeline0/GstQueue:queue0.GstPad:src: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
/GstPipeline:pipeline0/GstGamma:gamma0.GstPad:src: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
/GstPipeline:pipeline0/GstQueue:queue1.GstPad:sink: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
/GstPipeline:pipeline0/GstGamma:gamma0.GstPad:sink: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
/GstPipeline:pipeline0/GstQueue:queue1.GstPad:src: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
[INFO] [1737742649.174237272] [gst_image_sink_node]: preparing video with caps 'video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1'
[INFO] [1737742649.174275157] [gst_image_sink_node]: setcaps format string is RGB 
[INFO] [1737742649.174292274] [gst_image_sink_node]: setcaps n_components is 3
[INFO] [1737742649.174297083] [gst_image_sink_node]: setcaps bits is 8
[INFO] [1737742649.174301898] [gst_image_sink_node]: setcaps pixel_stride is 3
/GstPipeline:pipeline0/Rosimagesink:rosimagesink0.GstPad:sink: caps = video/x-raw, format=(string)RGB, height=(int)480, width=(int)640, framerate=(fraction)0/1
Redistribute latency...
[INFO] [1737742649.176075218] [gst_image_sink_node]: stream_start at 1737742649176073268
0:00:08.6 / 99:99:99.

After the pipeline stops, hitting Ctrl+C does not always stop GStreamer and I need to press Ctrl+\ twice in order to exit.

^\Caught SIGQUIT
Spinning.  Please run 'gdb gst-launch-1.0 81547' to continue debugging, Ctrl-C to quit, or Ctrl-\ to dump core.
^\Quit (core dumped)

I also tried running with FastDDS instead of CycloneDDS and saw a different failure mode, in which I receive an image only about every 10-20 seconds or so (much lower than the consistent ~30fps with rqt_image_view). I don't believe this is due to slow processing -- I'm streaming at only 480p on a 13th gen i9 laptop.

Any thoughts as to why this might be happening or how I could debug the issue further?