Deploying an Inference Container
Preparations
Before deploying container images, prepare the required container image software package. For details, see Step 4 in Creating an Inference Image.
Procedure
- Before deploying an inference container, add the SHA256 value of the container to the image trustlist by referring to Modify the security configurations of the AtlasEdge container.
- Deploy a container using FusionDirector. See Deploying Containers. The created image contains service inference programs. Therefore, you do not need to mount volumes and resource files when deploying a container on FusionDirector. The following is an example of creating an application template:
- (Optional) Input the service container data. The AtlasEdge container does not allow volumes to be mounted to the host. Therefore, you are advised to input data as follows:
- Input data by pulling streams, such as IPC streams, to devices from the edge container.
- Receive external data from the listening port in a container when port mapping is created during the container deployment.
- (Optional) Output the service container data. Output data in the service container scenario includes logs and inference results. The following output methods are recommended:
- Container logs and inference results can be output to external systems through the network, for example, to the log server and inference result application system.
- Container logs or inference results can be printed in standard output. Currently, they cannot be exported. You need to log in to the system to view them.
Parent topic: Creating and Deploying an Inference Container




