carla
carla copied to clipboard
CARLA 0.9.15 Docker container exits silently when using recorder in certain scenarios
Setup:
- official CARLA 0.9.15 image running inside Docker
- Client: carla_garage (local)
- GPU: NVIDIA A100
- No memory spikes observed during crash
Steps to Reproduce:
- Run CARLA 0.9.15 in a Docker container.
- Start a scenario with the client (e.g. BlockedIntersection or HighwayCutIn).
- Enable the recorder (/leaderboard_autopilot/leaderboard/leaderboard_evaluator_local.py, line 376: set argument=True).
Expected Behavior:
- Scenario runs with recording enabled.
Actual Behavior:
- As soon as the recorder starts, the Docker container exits without any error message.
- This only happens for some scenarios (BlockedIntersection, HighwayCutIn).
- Other scenarios (ParkingCutIn, Accident, ParkingCrossingPedestrian) run fine even with recorder.
- Without the recorder, all scenarios work as expected.
Notes:
- Ticking the simulation beforehand does not avoid the issue.
- The order of starting the recorder, loading the senario and running it does not matter.
- Appears to be more related to CARLA itself than carla_garage.
- No visible error logs in container, no GPU/CPU/SSD memory spikes.
I see the same behavior on CARLA 0.9.15 running natively. I do however get an error message, albeit essentially just a segfault. No carla_garage, i trigger this using the scenario runner.
orjahren@pensie7:~/Documents/master-extern/CARLA_0.9.15|⇒ ./CarlaUE4.sh
4.26.2-0+++UE4+Release-4.26 522 0
Disabling core dumps.
Signal 11 caught.
Malloc Size=65538 LargeMemoryPoolOffset=65554
CommonUnixCrashHandler: Signal=11
Malloc Size=131160 LargeMemoryPoolOffset=196744
Malloc Size=131160 LargeMemoryPoolOffset=327928
Engine crash handling finished; re-raising signal 11 for the default handler. Good bye.
./CarlaUE4.sh: line 5: 1056127 Segmentation fault (core dumped) "$UE4_PROJECT_ROOT/CarlaUE4/Binaries/Linux/CarlaUE4-Linux-Shipping" CarlaUE4 "$@"
My system details, if relevant
- GPU: RTX 4070
- OS: Fedora Linux 42
- Kernel: 6.16.7-200.fc42.x86_64
- GPU drivers: NVIDIA-SMI 580.82.09 Driver Version: 580.82.09 CUDA Version: 13.0
- Client: Scenario runner 0.9.15