Matthew Blau
01/22/2021, 2:58 PMZanie
Docker
storage for your flows and package them with your exiting code then use a docker agent to run them. Alternatively, you can use file based storage (e.g. S3 storage) for your flows and use your existing code as a base docker image. In that case, you’d set up a DockerRun
run config for your flow that runs it in your base image (or another run config that uses a docker image e.g. ECSRun
)Matthew Blau
01/22/2021, 3:19 PMZanie
Flows
then when you register the flow use Docker
storage with a reference your Dockerfile
extra_dockerfile_commands
)Matthew Blau
01/22/2021, 3:30 PMZanie
flow.register()
the default storage method is to take the flow object and pickle it on the local file system. When you then try to run your flow, it will not know that it should be executed in a docker container. This is why you need to specify flow.storage = Docker(…)
before registration. You can pass the path to your Dockerfile
in.Matthew Blau
01/22/2021, 3:43 PMZanie
Matthew Blau
01/22/2021, 3:55 PMZanie