An example of read module for Fybrik.
In this repository we show how to create a read module for Fybrik. We tested the read module with a python application that launches a web server to respond to GET requests of datasets.
- Kubernetes cluster 1.10+
- Helm 3.7.x and above
- Install Fybrik using the Quick Start guide.
- Docker repository (such as ghcr.io).
Create a file to implement your usage of the read module. An example can be found in hello-world-read-module.py
where you can find a python code that runs a simple web server and responds to GET requests of datasets.
In Makefile
:
- Create a registry for helm chart and docker image. Then change the fields
DOCKER_USERNAME
,DOCKER_PASSWORD
,DOCKER_HOSTNAME
,DOCKER_NAMESPACE
,DOCKER_TAGNAME
,DOCKER_NAME
, andHELM_TAG
to your own preferences. An example can be found inMakefile
. - One possible option is to create public registries in github. Then create a Personal Access Token. In this case the field
DOCKER_USERNAME
will be your github username andDOCKER_PASSWORD
is the Personal Access Token. Note that you need to change the visibility of the packages to public.
Create a Dockerfile to run your code that you created in a previous step. Then, run the following command to build a docker image from the Dockerfile.
make docker-build
Run the following command to login to the registry meant to store the docker image and to push the image to the registry.
make docker-push
This helm chart defines a common structure to deploy a Kubernetes pod for an Fybrik module. In the helm chart a service, a serviceaccount, and a deployment are defined.
- The helm chart defines some Kubernetes resources depending on the values in
values.yaml
. - Modify repository in
values.yaml
to your Docker image registry. - At runtime, the
fybrik-manager
will pass in the values (like data location, format, and credentials) to the module so you can leave them blank in your final chart.
Run the following command to login to the registry meant to store the helm chart.
make helm-login
Run the following command to create a helm chart from the helm directory hello-world-read-module
.
make helm-verify
Run the following command to login to your registry that intended to store the helm chart and to push the chart to the registry. Then, uninstall the helm chart.
make helm-chart-push
To register HWRM (Hello Read World Module) as a Fybrik module apply hello-world-read-module.yaml
to the fybrik-system namespace of your cluster.
In order to install the latest release, run:
kubectl apply -f https://github.com/fybrik/hello-world-read-module/releases/latest/download/hello-world-read-module.yaml -n fybrik-system
Fybrik | HWRM | Command |
---|---|---|
0.5.x | 0.5.x | https://github.com/fybrik/hello-world-read-module/releases/download/v0.5.0/hello-world-read-module.yaml |
0.6.x | 0.6.x | https://github.com/fybrik/hello-world-read-module/releases/download/v0.6.0/hello-world-read-module.yaml |
0.7.x | 0.7.x | https://github.com/fybrik/hello-world-read-module/releases/download/v0.7.0/hello-world-read-module.yaml |
master | main | https://raw.githubusercontent.com/fybrik/hello-world-read-module/main/hello-world-read-module.yaml |
Here is an example how to deploy and test the module on a single cluster.
Install Fybrik using the Quick Start guide. This sample assumes the use of the built-in catalog and Open Policy Agent (OPA).
Notice: Please follow
version compatbility matrix
section above for deploying the correct version of Fybrik and this module.
Deploy this module in the fybrik-system
namespace:
kubectl apply -f hello-world-read-module.yaml -n fybrik-system
Notice: Please use the README.md file of the desired release as the resources in this example may change between releases.
Execute the sections in Fybrik Notebook sample until Register the dataset in a data catalog
section (excluded).
You need to register your data asset in a data catalog in order for it to be used by the fybrik-manager
.
-
Follow step
Register the dataset in a data catalog
in this example. These steps register the credentials required for accessing the dataset, and then register the data asset in the catalog. -
As an example you can run these commands to register two assets exist in
sample_assets
:
kubectl apply -f sample_assets/assetMedals.yaml -n fybrik-notebook-sample
kubectl apply -f sample_assets/secretMedals.yaml -n fybrik-notebook-sample
kubectl apply -f sample_assets/assetBank.yaml -n fybrik-notebook-sample
kubectl apply -f sample_assets/secretBank.yaml -n fybrik-notebook-sample
Define the following OpenPolicyAgent policy to allow the write operation:
package dataapi.authz
rule[{"action": {"name":"RedactAction", "columns": column_names}, "policy": description}] {
description := "Redact columns tagged as PII in datasets tagged with finance = true"
input.action.actionType == "read"
input.resource.metadata.tags.finance
column_names := [input.resource.metadata.columns[i].name | input.resource.metadata.columns[i].tags.PII]
count(column_names) > 0
}
rule[{"action": {"name":"RedactAction", "columns": column_names}, "policy": description}] {
description := "Redact columns tagged as sensitive in datasets tagged with finance = true"
input.action.actionType == "read"
input.resource.metadata.tags.finance
column_names := [input.resource.metadata.columns[i].name | input.resource.metadata.columns[i].tags.sensitive]
count(column_names) > 0
}
Copy the policies to a file named sample-policy.rego and then run:
kubectl -n fybrik-system create configmap sample-policy --from-file=sample-policy.rego
kubectl -n fybrik-system label configmap sample-policy openpolicyagent.org/policy=rego
while [[ $(kubectl get cm sample-policy -n fybrik-system -o 'jsonpath={.metadata.annotations.openpolicyagent\.org/policy-status}') != '{"status":"ok"}' ]]; do echo "waiting for policy to be applied" && sleep 5; done
Deploy FybrikApplication
in default
namespace:
kubectl apply -f fybrikapplication.yaml -n default
- Run the following command to wait until the
status
of theFybrikApplication
isready
:
while [[ $(kubectl get fybrikapplication my-notebook -n default -o 'jsonpath={.status.ready}') != "true" ]]; do echo "waiting for FybrikApplication" && sleep 5; done
- Check if module was triggered in
fybrik-blueprints
:
kubectl get blueprint -n fybrik-system
kubectl describe blueprint my-notebook-default -n fybrik-system
kubectl get job -n fybrik-blueprints
kubectl get pods -n fybrik-blueprints
If you are using the hello-world-read-module
image, you should see this in the kubectl logs -n fybrik-blueprints
of your completed Pod:
$ kubectl logs my-notebook-default-hello-world-read-module-xx -n fybrik-blueprints
INFO:root:
Hello World Read Module!
INFO:root:The avialable datasets:
INFO:root:dataset name: medals-winners
INFO:root: format: csv
INFO:root: endpoint_url: http://winterolympicsmedals.com/medals.csv
INFO:root: action: Redact
INFO:root: transferred_columns: ['age']
INFO:root:dataset name: bank
INFO:root: format: csv
INFO:root: endpoint_url: https://raw.githubusercontent.com/juliencohensolal/BankMarketing/master/rawData/bank-additional-full.csv
INFO:root: action: Redact
INFO:root: transferred_columns: ['age']
INFO:root:Starting httpd server on localhost:8000
Run the following command to delete the fybrik application:
kubectl delete FybrikApplication my-notebook -n default
Run the following command to delete the fybrik module:
kubectl delete fybrikmodule hello-world-read-module -n fybrik-system
Please execute the Cleanup
section from Fybrik notebook sample