docker-backend

Containerization of the DeepHealth back end (see https://github.com/deephealthproject/backend).

Keep on reading to see how to deploy on Kubernetes or how to run it or develop using docker-compose.

Deploy on Kubernetes

Requirements

• Kubernetes volume provisioner with support for ReadWriteMany access mode (e.g., Ceph, NFS, etc.)

Installation

1. Add the helm-charts repository:

helm repo add dhealth https://deephealthproject.github.io/helm-charts/
helm repo update

2. Download the values.yaml template:

curl https://raw.githubusercontent.com/deephealthproject/docker-backend/develop/k8s/deephealth-backend/values.yaml -o values.yaml

3. Edit the values.yaml to configure your deployment (here the available parameters)
4. To install the chart with the release name deephealth-backend :

helm install --name deephealth-backend dhealth/deephealth-backend -f values.yaml

Take a look at the deployment notes to know how to access services (helm status deephealth-backend).

Installation from sources

1. Clone the docker-backend repository:

git clone [email protected]:deephealthproject/docker-backend.git

2. Collect helm dependencies by typing:

helm repo add bitnami https://charts.bitnami.com/bitnami
helm repo add crs4 https://crs4.github.io/helm-charts/
helm repo add deephealth https://deephealthproject.github.io/helm-charts/
helm dependency build k8s/deephealth-backend

3. Edit the ´k8s/values.yaml´ template to configure your deployment (here the available parameters)

4. Deploy on Kubernetes through helm:

helm install --name deephealth-backend k8s/deephealth-backend -f k8s/values.yaml

Take a look at the deployment notes to know how to access services (helm status deephealth-backend).

Access via Ingress

If an Ingress Controller is available on your Kubernetes cluster, you can access the back-end endpoint by setting the upper level ingress property on the values.yaml file. Here is an example:

ingress:
  enabled: true
  annotations:
    kubernetes.io/ingress.class: nginx
    # kubernetes.io/tls-acme: "true"
  hosts:
      # Replace host with your
      # ingress controller hostname or IP
    - host: backend.172.30.20.10.nip.io

Parameters

The following tables lists the main configurable parameters of the deephealth-backend chart and their default values.

Parameter	Description	Default
global.debug	Enable/disable debug mode	False
global.storageClass	Global storage class for dynamic provisioning	nil (i.e., the default sc in your cluster)
global.imagePullPolicy	Global image pull policy	Always
global.retainSecrets	Preserve secrets when a release is deleted	True
global.retainPVCs	Preserve PVCs when a release is deleted	True
endpoint.service.type	Kubernetes service type of the API Endpoint	NodePort
ingress.enabled	Enable the ingress for the API Endpoint service	false
ingress.annations	Annotations for the ingress realated with the API Endpoint service	kubernetes.io/ingress.class: nginx
ingress.hosts	Hosts paths for the ingress realated with the API Endpoint service (see example)	nil
backend.image.repository	Back-end App Docker Image	dhealth/backend
backend.image.tag	Back-end App Docker Image Tag	2fd828d8
backend.admin.username	Username of the administrator of the backend app	admin
backend.admin.password	Password of the administrator of the backend app (autogenerated if not defined)	nil
backend.admin.password	Password of the administrator of the backend app (autogenerated if not defined)	nil
backend.social_auth.github.client_id	ClientID of the back-end registered as OAuth2 app on Github	nil
backend.social_auth.github.client_secret	Client Secret of the back-end registered as OAuth2 app on Github	nil
backend.email.user	Username for the the back-end email service	nil
backend.email.password	Password for the the back-end email service	nil
backend.replicaCount	Number of replicase of the the backend (Gunicorn) server replicas	1
backend.workers	Number of workers of every backend (Gunicorn) server replicas	3
backend.persistence.data.storageClass	Storage class used for backend data (requires support for ReadWriteMany access mode)	*globalStorageClass
backend.persistence.data.path	Path to mount the backend data volume at	/data
backend.persistence.data.size	Size of the backend data volume (i.e., datasets, inference and training data)	10Gi
backend.allowedHosts	List of ALLOWED_HOSTS for the backend Django app	*
backend.corsOriginWhiteList	CORS_ORIGIN_WHITE_LIST for the backend Django app	nil
backend.resources	CPU/Memory resource requests/limits of the backend server replicas	nil
backend.nodeSelector	Node labels for pod assignment of the backend server replicas	nil
backend.tolerations	Tolerations labels for pod assignment of the backend server repliacas	nil
backend.affinity	Affinity labels for pod assignment of the backend server replicas	nil
backend.podLabels	Extra labels for backend pods	nil
backend.podSecurityContext	Backend pods' security context	nil
celery.acceptContent	A list of comma-separated content-types to allow on Celery workers	json
celery.taskSerializer	A list of comma-separated serializers to allow on Celery workers	json
celery.resources	CPU/Memory resource requests/limits of the celery worker replicas	nil
celery.nodeSelector	Node labels for pod assignment of the celery worker replicas	nil
celery.tolerations	Tolerations labels for pod assignment of the celery worker replicas	nil
celery.affinity	Affinity labels for pod assignment of the celery worker replicas	nil
celery.podLabels	Extra labels for celery pods	nil
celery.podSecurityContext	Celery pods' security context	nil
broker.persistence.storageClass	Storage class used by RabbitMQ broker	*globalStorageClass
broker.persistence.size	Size of the datasets volume	1Gi
postgresql.postgresqlPostgresPassword	PostgreSQL admin password (used when postgresqlUsername is not postgres)	random 10 character alphanumeric string
postgresql.postgresqlUsername	PostgreSQL admin user	postgres
postgresql.postgresqlPassword	PostgreSQL admin password	random 10 character alphanumeric string
postgresql.postgresqlDatabase	PostgreSQL database	db
postgresql.persistence.storageClass	Storage class for PostgreSQL volume	*globalStorageClass
postgresql.persistence.size	Storage size of PostgreSQL volume	1Gi

Below you can find the link to the available parameters for the remaining deployment components:

• nginx: https://github.com/bitnami/charts/blob/master/bitnami/nginx/README.md
• broker: https://github.com/bitnami/charts/blob/master/bitnami/rabbitmq/README.md
• postgresql: https://github.com/bitnami/charts/blob/master/bitnami/postgresql/README.md

Develop with Docker(Compose)

docker-backend allows to easily setup a development environment.

Requirements

• Docker (>= 19.03.5)
• DockerCompose (>=1.25.4)

Installation

1. Edit the file settings.conf to customize your setup (e.g., services credentials and ports).

   Some relevant settings are:

   • DOCKER_LIBS_IMAGE, i.e., the docker-libs image to use (see deephealthproject/docker-libs), important to ensure the support of specific versions of the DeepHealth libraries;

   • DOCKER_RUNTIME, i.e., the default Docker runtime runc or nvidia to leverage GPUs;

   • BACKEND_LOCAL_PATH, to mount and directly develop with your local copy of the deephealthproject/backend on Docker

2. Type ./setup.sh to generate configuration files and generate the Docker images.

Usage

Type docker-compose up -d to start all the services and after a while you should have the Back-end API up and running at http://<YOUR_HOST_IP>:<BACKEND_PORT> (http://localhost:8000 by default).

Use docker-compose logs [service] to see the logs of the running services.

Type docker-compose down to stop and remove all the containers in use.

License

This software is distributed under the MIT License, see LICENSE for more information.