Carla: [0.9.7] Loosing connection to remote Carla instance

[semmelknoedel]

I'm running a docker container of version 0.9.7 on a remote machine. When I start manual_control.py there it works fine.
When I start manual_control.py on my local machine I get the following output:

```
/usr/bin/python3.6 /path/to/carla/PythonAPI/examples/manual_control.py --host x.x.x.x
pygame 1.9.6
Hello from the pygame community. https://www.pygame.org/contribute.html

Welcome to CARLA manual control.

Use ARROWS or WASD keys for control.

W            : throttle
S            : brake
AD           : steer
Q            : toggle reverse
Space        : hand-brake
P            : toggle autopilot
M            : toggle manual transmission
,/.          : gear up/down

TAB          : change sensor position
`            : next sensor
[1-9]        : change to sensor [1-9]
G            : toggle radar visualization
C            : change weather (Shift+C reverse)
Backspace    : change vehicle

R            : toggle recording images to disk

CTRL + R     : toggle recording of simulation (replacing any previous)
CTRL + P     : start replaying last recorded simulation
CTRL + +     : increments the start time of the replay by 1 second (+SHIFT = 10 seconds)
CTRL + -     : decrements the start time of the replay by 1 second (+SHIFT = 10 seconds)

F1           : toggle HUD
H/?          : toggle help
ESC          : quit

INFO: listening to server x.x.x.x:2000
WARNING: Version mismatch detected: You are trying to connect to a simulator that might be incompatible with this API
WARNING: Client API version = 0.9.7-41-g68d7418e
WARNING: Simulator API version = 0.9.7
No recommended values for 'speed' attribute
Traceback (most recent call last):
File "/path/to/carla/PythonAPI/examples/manual_control.py", line 1034, in
main()
File "/path/to/carla/PythonAPI/examples/manual_control.py", line 1026, in main
game_loop(args)
File "/path/to/carla/PythonAPI/examples/manual_control.py", line 943, in game_loop
controller = KeyboardControl(world, args.autopilot)
File "/path/to/carla/PythonAPI/examples/manual_control.py", line 276, in __init__
world.player.set_autopilot(self._autopilot_enabled)
RuntimeError: time-out of 5000ms while waiting for the simulator, make sure the simulator is ready and connected to localhost:2000
Fatal Python error: PyEval_SaveThread: NULL tstate

Process finished with exit code 134 (interrupted by signal 6: SIGABRT)

```

I have the same problem with spawn_npc.py. But it actually spawns some npcs at the remote machine before the connection crashes (so at the beginning there is a connection). Also I am confused that the last error message is showing localhost:2000 but I clearly gave an address.

Both, remote and locale machine, are running with Ubuntu 18.04 LTS.

Has someone seen this problem before?

[pawel-ziecina]

I'm observing similar issue on local machine. Steps to reproduce:

1. Start original carla 0.9.7.3 release with

./CarlaUE4.sh -opengl -carla-rpc-port=36540 -carla-streaming-port=0 -quality-level=Low -nosound

1. Check if carla server properly binded

netstat -nlp|grep -i carla

tcp        0      0 0.0.0.0:44581           0.0.0.0:*               LISTEN      21795/CarlaUE4-Linu 
tcp        0      0 0.0.0.0:36540           0.0.0.0:*               LISTEN      21795/CarlaUE4-Linu 

1. Run example spawn_npc.py script:

PYTHONPATH=$PWD/PythonAPI/carla/dist/carla-0.9.7-py3.5-linux-x86_64.egg python3.6 ./PythonAPI/examples/spawn_npc.py -p 36540 --host 127.0.0.1

Observing:

destroying 0 vehicles

destroying 0 walkers

done.
Traceback (most recent call last):
  File "./PythonAPI/examples/spawn_npc.py", line 226, in <module>
    main()
  File "./PythonAPI/examples/spawn_npc.py", line 124, in main
    for response in client.apply_batch_sync(batch):
RuntimeError: time-out of 5000ms while waiting for the simulator, make sure the simulator is ready and connected to localhost:2000
Fatal Python error: PyEval_SaveThread: NULL tstate

On default port on localhost it works correctly.

[semmelknoedel]

As a temporal solution I'm using port forwarding right now:
ssh -L 2000:localhost:2000 -L 2001:localhost:2001 -L 2002:localhost:2002 user@remote_host

[jackbart94]

Hi,

A solution to this issue will be in master very soon, thanks for pointing it out!

[liespace]

@jackbart94 Has this bug been fixed? I use the latest version docker but still meet this bug.

[jackbart94]

@liespace It's being merged in master as we speak!

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.