transformers/

README

Transformers

We have sample transformers echo, md5 and tar2tf.

Transformer	Language	Description
echo	python:3.8.5	This returns the original data, with an md5 sum in the response headers.
go-echo	golang:1.15	This returns the original data, with an md5 sum in the response headers.
hello_world	python:3.8.5	This always returns Hello World! string on any request.
md5	python:3.8.5	This returns the md5 sum of the original data as the response.
tar2tf	golang:1.13	This returns the transformed TensorFlow compatible data for the input tar files.

Each of sample transformers contains three main files:

• Dockerfile - contains description of Docker image.
• pod.yaml - contains the spec for the K8s Pod running this transformer.
• Makefile - used for building and pushing Docker image.

Using sample transformers

Environment Variables

Env Variable	Default Value	Description
${COMMUNICATION_TYPE}	"hpush://"	This is the mode of communication to be used. For more info read this.

Terms

HostMachine - The place from where the deployment is performed.

Prerequisites

1. kubectl must be installed on HostMachine and connected and correctly configured for aistore kubernetes network.
2. docker must be installed on the HostMachine.

Steps:

1. We must set the ${COMMUNICATION_TYPE} environment variable first with appropriate value.
2. eval "echo \"$(cat pod.yaml)\"" > pod_config.yaml - substitutes the environment variables
3. make build - builds the Docker image
4. docker push <IMAGE_URL> - <IMAGE_URL> can be obtained post the build step
5. ais transform init pod_config.yaml