aws-container-launcher 0.0.1

AWS Container Launcher helps to run a docker container in AWS

Prerequisites

• Python 3.7
• AWS Credentials

AWS Container Launcher (aws-container-launcher)

AWS Container Launcher helps to run a docker container in AWS.

Examples

WARNING: After processing a job, you are in charge of destroying the infrastructure. Do not forget to call the "destroy" method; otherwise, the instance will continue to operate and you will be charged by Amazon.

Minimal example with providing an instance type

import time
from aws_container_launcher import ContainerLauncher, RunStatus

acl = ContainerLauncher()

start_container_input = {
    'commands': [
        'sleep 1'
    ],  # required
    'docker_image': 'ubuntu:latest',  # required
    'subnet_ids': [<subnet_id_1>, <subnet_id_2>],  # required
    'instance_type': 't3.micro',  # required. Alternatively, use 'cpu' and 'memory' parameters pair.
}

'''
Asynchronously running of an ECS task. However, resource initializing happens synchronously (EC2, etc.) and takes up to 5 minutes
'''
start_container_response = acl.start_container(start_container_input)

entry_id = start_container_response['entry_id']

time.sleep(10)

get_container_status_response = acl.get_container_status(entry_id)

status = get_container_status_response['status']

assert status == RunStatus.COMPLETED.name

acl.destroy(entry_id)

Minimal example with providing 'cpu' and 'memory' pair

import time
from aws_container_launcher import ContainerLauncher, RunStatus

acl = ContainerLauncher()

start_container_input = {
    'commands': [
        'sleep 1'
    ],  # required
    'docker_image': 'ubuntu:latest',  # required
    'subnet_ids': [<subnet_id_1>, <subnet_id_2>],  # required
    'cpu': 1,  # required. Alternatively, use the 'instance_type' parameter for 'cpu' and 'memory' parameters pair.
    'memory': 4  # required. Alternatively, use the 'instance_type' parameter for 'cpu' and 'memory' parameters pair.
}

'''
Asynchronously running of an ECS task. However, resource initializing happens synchronously (EC2, etc.) and takes up to 5 minutes
'''
start_container_response = acl.start_container(start_container_input)

entry_id = start_container_response['entry_id']

time.sleep(10)

get_container_status_response = acl.get_container_status(entry_id)

status = get_container_status_response['status']

assert status == RunStatus.COMPLETED.name

acl.destroy(entry_id)

Advanced

import time
import logging
from aws_container_launcher import ContainerLauncher, RunStatus

logger = logging
logging.basicConfig(level=logging.INFO)

acl = ContainerLauncher({
    'monitoring_table_tags': {
        'k1': 'v1',
        'k2': 'v2'
    },  # optional. It is tagging for the monitoring DynamoDB table.
        # The DynamoDB table will be created if it doesn't exist and will be reused if it exists.
    'logger': logger,
    'region': 'us-east-1'
})

entry_id = str(time.time()).replace('.', '')

start_container_input = {
    'commands': [
        'sleep 1'
    ],  # required
    'docker_image': 'ubuntu:latest',  # required
    'subnet_ids': [<subnet_id_1>, <subnet_id_2>],  # required
    'cpu': 1,  # required. Alternatively, use the 'instance_type' parameter for 'cpu' and 'memory' parameters pair.
    'memory': 4,  # required. Alternatively, use the 'instance_type' parameter for 'cpu' and 'memory' parameters pair.
    'id': entry_id,  # optional. The ID must be unique for each task.
                     # Otherwise, you will get the same status for an already completed job.
                     # If not provided, then ID will be generated automatically.
    'ec2_sg_id': <security_group_id>,  # optional. Default: A default VPC security group
    'tags': {
        'k1': 'v1',
        'k2': 'v2'
    },  # optional
    'is_spot': False,  # Default: True
    'instance_type': 't3.micro',  # required. Alternatively, use 'cpu' and 'memory' parameters pair.
                                     # The instance type has priority over 'cpu' and 'memory' parameters pair.
    'storage': {
        'type': 'fsx',  # fsx support only
        'file_system_id': <file_system_id>,
        'dir': '/fsx'  # It is a path to access FSx space from a docker container. Default: /fsx
    },  # optional
    'options': {
        'scaling_enabled': True
    }  # Used as a safety catch if you forgot to destroy the infrastructure after processing the job.
       # If the CPU activity for the EC2 instance is less than 1 percent for 3 hours,
       # the EC2 instance will be terminated by Auto Scaling Policy.
}

'''
Asynchronously running of an ECS task.
However, resource initializing happens synchronously (EC2, etc.)
and takes up to 5 minutes
'''
start_container_response = acl.start_container(start_container_input)

entry_id = start_container_response['entry_id']

time.sleep(10)

get_container_status_response = acl.get_container_status(entry_id)

status = get_container_status_response['status']

print(status)

assert status == RunStatus.COMPLETED.name

acl.destroy(entry_id)

Storage

Amazon FSx for Lustre

start_container_input = {
    'commands': [
        'sleep 1'
    ],
    'docker_image': 'ubuntu:latest',
    'subnet_ids': [<subnet_id_1>, <subnet_id_2>],
    'cpu': 1,
    'memory': 4,
    'storage': {
        'type': 'fsx',
        'file_system_id': <file_system_id>,
        'dir': '/fsx'
    }
}

Security

Currently, all IAM roles have policies with administrator rights. The roles will eventually be adjustable in the future.