This document describes common use cases (building, testing, extracting) for the Data Processing Docker Image
Note: At this time, FB does not allow docker builds on FB devservers. In order to build data_processing docker image(s) you will need to either clone fbcode locally to a docker capable machine (your laptop) or at a minimum copy the "data_processing" folder to a machine that is capable of using docker.
Data Processing docker image uses the fbpcf docker image as its base. We currently do not have a public docker registry, so this will need to be build locally first. You have two options to build this image...
- Option 1: Clone https://github.com/facebookresearch/fbpcf from github and simply run
build_docker.sh
in the root of the project - Option 2: (If you cloned fbcode locally) Goto fbsource/fbcode/measurement/private_measurement/oss and simply run "build_docker.sh" in the root of the project
Note: If building locally you may need to increase the docker memory resources to 4GB in the preferences
To build data processing and data_processing:latest
docker image run the following script
- `./build-docker.sh
- build-docker currently only supports Ubuntu but we might support Alpine in the future
Simply run docker/run-attribution_id_combiner-test.sh
and output will be stored in test-output
folder (created automatically)
To upload to a new data_processing docker image to AWS
Prerequisite: aws cli must be installed
- MacOS:
brew install awscli
Upload/update in AWS:
aws ecr get-login-password --region us-west-2 | docker login --username AWS --password-stdin 539290649537.dkr.ecr.us-west-2.amazonaws.com
docker tag data_processing:latest 539290649537.dkr.ecr.us-west-2.amazonaws.com/data-processing:latest
docker push 539290649537.dkr.ecr.us-west-2.amazonaws.com/data-processing:latest