nodelets running with remote roscore are dying (bond broken)
Hi all (or, at least, the usual one who answers to my questions ;-)), I have the following situation:
- LAPTOP1: roscore, roslaunch with openni_node and some tf
- LAPTOP2: roslaunch with some depth_image_proc nodelets, connected via ROS_MASTER_URI=http://LAPTOP1
the problem is that the nodelets on LAPTOP2 are alive only for a short while (some seconds), then, they start to die, one after the other:
[ INFO] [1396535717.713638183]: Bond broken, exiting
[camera/register_depth_rgb-3] process has finished cleanly
What's going on? How can I run this configuration properly? Thanks
I tested also what happens without the **openni_node** on LAPTOP1 (i.e., there is no input for the nodelets on LAPTOP2, so they should not run, since their subscribed topics are not published): same behavior
What happens when you start the roscore on laptop2 instead?
Moreover, do other nodes / nodelets have the same problems on laptop2, or is it just the image_proc nodelets?
I can run the nodelets on laptop2 with a roscore on laptop2, everything works (this is actually how everything has begun). Can you tell me a simple nodelet I can test (to see if it's a depth_image_proc problem or nodelet problem)?
Have you tried simple nodes on laptop2 with core on laptop1? Something like `rostopic pub` or `rostopic sub`?
The openni_node running on laptop1 publishes /camera/rgb/image_raw and /camera/depth/image_raw (roscore is on laptop1), I can see that image by subscribing it from laptop2 with rqt_image_view.
Ok. I don't know what could be the problem, sorry. @esteve, @tfoote, any idea how to investigate this best?
I rewrote launch files, rebooted laptops, upgraded packages... now it works (more or less)... could it be that was because I had image_transport package installed on one laptop (LAPTOP1) and not the other?