I am planning to use an nvidia jetson tx2 as drone server/agent.
So are there any known issues running drone
- on arm?
- while starting with the nvidia runtime (https://github.com/NVIDIA/nvidia-docker)?
I am planning to use an nvidia jetson tx2 as drone server/agent.
So are there any known issues running drone
Yes, Drone fully supports arm and arm64 architectures (for example, the Alpine Linux project uses Drone to test packages on arm and arm64). Unfortunately I have no experience with nvidia so I cannot answer that question.
Sorry for digging out an old thread.
We are facing the same challenge. We successfully run a Drone.io agent on an arm64 Nvidia Jetson TX2 board. Without the --runtime=nvidia option so far. We’ll add a Nvidia Xavier later to it.
As we couldn’t find a drone pipeline parameter to specify the docker container’s runtime, we found that in /etc/docker/daemon.json you can specify the default runtime. So in our case running all containers in the nvidia runtime is acceptable so we could specify it as default. This could do the trick for you as well.
(https://docs.docker.com/engine/reference/commandline/dockerd/#docker-runtime-execution-options)
In the meantime we have to fix our broken nvidia runtime on the TX2 to do the testing.