Awesome
fargatespawner
Spawns JupyterHub single user notebook servers in Docker containers running in AWS Fargate
Installation
pip install fargatespawner
Configuration
To configure JupyterHub to use FargateSpawner, you can add the following to your jupyterhub_config.py
.
from fargatespawner import FargateSpawner
c.JupyterHub.spawner_class = FargateSpawner
You must also set the following settings on c.FargateSpawner
in your jupyterhub_config.py
. None of them are optional.
Setting | Description | Example |
---|---|---|
aws_region | The AWS region in which the tasks are launched. | 'eu-west-1' |
aws_ecs_host | The hostname of the AWS ECS API. Typically, this is of the form ecs.<aws-region>.amazonaws.com . | 'ecs.eu-west-1.amazonaws.com' |
notebook_port | The port the notebook servers listen on. | 8888 |
notebook_scheme | The scheme used by the hub and proxy to connect to the notebook servers. At the time of writing 'https' will not work out of the box. However, users do not connect to the the notebook server directly, and does not, typically, allow incoming connections from the public internet. Instead, users connect to the proxy, which can be configured to listen on HTTPS independently of this setting. There is more information on setting up HTTPS for connections to the proxy at https://jupyterhub.readthedocs.io/en/stable/getting-started/security-basics.html. | 'http' |
get_run_task_args | See below | See below |
The get_run_task_args
argument must a callable that takes the spawner instance as a parameter, and returns a dictionary that is passed to the RunTask
API call.
c.FargateSpawner.get_run_task_args = lambda spawner: {
'cluster': 'jupyerhub-notebooks',
'taskDefinition': 'jupyterhub-notebook:7',
'overrides': {
'taskRoleArn': 'arn:aws:iam::123456789012:role/notebook-task',
'containerOverrides': [{
'command': spawner.cmd + [f'--port={spawner.notebook_port}', '--config=notebook_config.py'],
'environment': [
{
'name': name,
'value': value,
} for name, value in spawner.get_env().items()
],
'name': 'jupyterhub-notebook',
}],
},
'count': 1,
'launchType': 'FARGATE',
'networkConfiguration': {
'awsvpcConfiguration': {
'assignPublicIp': 'DISABLED',
'securityGroups': ['sg-00026fc201a4e374b'],
'subnets': ['subnet-01fc5f15ac710c012'],
},
},
}
The parts of the return value that probably require change from the above example are:
-
cluster - The name of the ECS cluster into which the tasks are launched.
-
taskDefinition - The family:revision or full ARN of the task definition that runs the notebooks. Typically, this task definition would specify a docker image that builds on one from https://github.com/jupyter/docker-stacks.
-
taskRoleArn - The role the notebook tasks can assume. For example, in order for them to make requests to AWS, such as to use Jupyter S3 with role-based authentication.
-
securityGroups - The security groups associated with the Fargate task. They must allow communication to and from the hub/proxy. More information, such as the ports used, is at https://jupyterhub.readthedocs.io/en/stable/getting-started/networking-basics.html.
-
subnets - The list of possible subnets in which the task will start.
You must also, either, authenticate using a secret key, in which case you must have the following configuration
from fargatespawner import FargateSpawnerSecretAccessKeyAuthentication
c.FargateSpawner.authentication_class = FargateSpawnerSecretAccessKeyAuthentication
and the following settings on c.FargateSpawnerSecretAccessKeyAuthentication
Setting | Description | Example |
---|---|---|
aws_access_key_id | The ID of the AWS access key used to sign the requests to the AWS ECS API. | ommitted |
aws_secret_access_key | The secret part of the AWS access key used to sign the requests to the AWS ECS API . | ommitted |
or authenticate using a role in an ECS container, in which case you must have the following configuration
from fargatespawner import FargateSpawnerECSRoleAuthentication
c.FargateSpawner.authentication_class = FargateSpawnerECSRoleAuthentication
where FargateSpawnerECSRoleAuthentication does not have configurable options.
or authenticate using a IAM profile attached to EC2 instance, in which case you must have the following configuration:
from fargatespawner import FargateSpawnerEC2InstanceProfileAuthentication
c.FargateSpawner.authentication_class = FargateSpawnerEC2InstanceProfileAuthentication
Run-time dependencies
The spawner is deliberately written to not have any additional dependencies, beyond those that are required for JupyterHub.
Approximate minimum permissions
In order for the user to be able to start, monitor, and stop the tasks, they should have the below permissions.
{
"Version": "2012-10-17",
"Statement": [
{
"Sid": "",
"Effect": "Allow",
"Action": "ecs:RunTask",
"Resource": "arn:aws:ecs:<aws_region>:<aws_account_id>:task-definition/<task_family>:*",
"Condition": {
"ArnEquals": {
"ecs:cluster": "arn:aws:ecs:<aws_region>:<aws_account_id>:cluster/<cluster_name>"
}
}
},
{
"Sid": "",
"Effect": "Allow",
"Action": "ecs:StopTask",
"Resource": "arn:aws:ecs:<aws_region>:<aws_account_id>:task/*",
"Condition": {
"ArnEquals": {
"ecs:cluster": "arn:aws:ecs:<aws_region>:<aws_account_id>:cluster/<cluster_name>"
}
}
},
{
"Sid": "",
"Effect": "Allow",
"Action": "ecs:DescribeTasks",
"Resource": "arn:aws:ecs:<aws_region>:<aws_account_id>:task/*",
"Condition": {
"ArnEquals": {
"ecs:cluster": "arn:aws:ecs:<aws_region>:<aws_account_id>:cluster/<cluster_name>"
}
}
},
{
"Sid": "",
"Effect": "Allow",
"Action": "iam:PassRole",
"Resource": [
"arn:aws:iam::<aws_account_id>:role/<task-execution-role>",
"arn:aws:iam::<aws_account_id>:role/<task-role>"
]
}
]
}