This PR integrates GCP CloudRun to deploy to the cloud.
For the detailed setup, please refer to the README
By requesting the /start request (status_request is for the status query),
the corresponding GCP container service will be spawned.
The service will have an automatic domain for the service. For the following example, it is https://hello-gol57lmytq-du.a.run.app. Accessing it will serve a test page made by Google.
Stopping the service will delete the GCP container service.
NOTE
GPU is not currently supported by GCP CloudRun for container deploy. We may need to use Anthos to make GPU attachable in the GCP container service.
NOTE
Currently, the deployable service is limited to the default example of GCP service (refer to example deployment YAML). PyTorch inference service scenarios are soon to be developed.
This PR integrates GCP CloudRun to deploy to the cloud.
For the detailed setup, please refer to the README
By requesting the
/start
request (status_request
is for the status query),the corresponding GCP container service will be spawned.
The service will have an automatic domain for the service. For the following example, it is https://hello-gol57lmytq-du.a.run.app. Accessing it will serve a test page made by Google.
Stopping the service will delete the GCP container service.