Running same Docker image with different flags on AWS and saving results - amazon-web-services

Question

I have an experiment I'd like to run 100 different times, each with a command line flag set to a different integer value. Each experiment will output the result to a text file. Experiments take about 2 hours each and are independent of each other.
I currently have a Docker image that can run the experiment when provided the command line flag.
I am curious if there is a way to write a script that can launch 100 AWS instances (one for each possible flag value), run the Docker image, and then output the result to a shared text file somewhere. Is this possible? I am very inexperienced with AWS so I'm not sure if this is the proper tool or what steps would be required (besides building the Docker image).
Thanks.

Answer 1

You could do this using vagrant with the vagrant-aws plugin to spin up the instances and the Docker Provisioner to pull your images / run your containers or the Ansible Provisioner. For example:
.
├── playbook.yml
└── Vagrantfile
The Vagrantfile:
# -*- mode: ruby -*-
# vi: set ft=ruby :
VAGRANTFILE_API_VERSION = "2"
Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
N = 100
(1..N).each do |server_id|
config.vm.box = "dummy"
config.ssh.forward_agent = true
config.vm.define "server#{server_id}" do |server|
server.vm.provider :aws do |aws, override|
aws.access_key_id = ENV["AWS_ACCESS_KEY_ID"]
aws.secret_access_key = ENV["AWS_SECRET_ACCESS_KEY"]
aws.instance_type = "t2.micro"
aws.block_device_mapping = [
{
"DeviceName" => "/dev/sda1",
"Ebs.VolumeSize" => 30
}
]
aws.tags = {
"Name" => "node#{server_id}.example.com",
"Environment" => "stage"
}
aws.subnet_id = "subnet-d65893b0"
aws.security_groups = [
"sg-deadbeef"
]
aws.region = "eu-west-1"
aws.region_config "eu-west-1" do |region|
region.ami = "ami-0635ad49b5839867c"
region.keypair_name = "ubuntu"
end
aws.monitoring = true
aws.associate_public_ip = false
aws.ssh_host_attribute = :private_ip_address
override.ssh.username = "ubuntu"
override.ssh.private_key_path = ENV["HOME"] + "/.ssh/id_rsa"
override.ssh.forward_agent = true
end
if server_id == N
server.vm.provision :ansible do |ansible|
ansible.limit = "all"
ansible.playbook = "playbook.yml"
ansible.compatibility_mode = "2.0"
ansible.raw_ssh_args = "-o ForwardAgent=yes"
ansible.extra_vars = {
"ansible_python_interpreter": "/usr/bin/python3"
}
end
end
end
end
end
Note: this example does ansible parallel execution from the Tips & Tricks.
The ansible playbook.yml:
- hosts: all
pre_tasks:
- name: get instance facts
local_action:
module: ec2_instance_facts
filters:
private-dns-name: '{{ ansible_fqdn }}'
"tag:Environment": stage
register: _ec2_instance_facts
- name: add route53 entry
local_action:
module: route53
state: present
private_zone: yes
zone: 'example.com'
record: '{{ _ec2_instance_facts.instances[0].tags["Name"] }}'
type: A
ttl: 7200
value: '{{ _ec2_instance_facts.instances[0].private_ip_address }}'
wait: yes
overwrite: yes
tasks:
- name: install build requirements
apt:
name: ['python3-pip', 'python3-socks', 'git']
state: present
update_cache: yes
become: true
- name: apt install docker requirements
apt:
name: ['apt-transport-https', 'ca-certificates', 'curl', 'gnupg-agent', 'software-properties-common']
state: present
become: true
- name: add docker apt key
apt_key:
url: https://download.docker.com/linux/ubuntu/gpg
state: present
become: true
- name: add docker apt repository
apt_repository:
repo: 'deb [arch=amd64] https://download.docker.com/linux/ubuntu bionic stable'
state: present
become: true
- name: apt install docker-ce
apt:
name: ['docker-ce', 'docker-ce-cli', 'containerd.io']
state: present
update_cache: yes
become: true
- name: get docker-compose
get_url:
url: 'https://github.com/docker/compose/releases/download/1.24.1/docker-compose-{{ ansible_system }}-{{ ansible_userspace_architecture }}'
dest: /usr/local/bin/docker-compose
mode: '0755'
become: true
- name: pip install docker and boto3
pip:
name: ['boto3', 'docker', 'docker-compose']
executable: pip3
- name: create docker config directory
file:
path: /etc/docker
state: directory
become: true
- name: copy docker daemon.json
copy:
content: |
{
"group": "docker",
"log-driver": "journald",
"live-restore": true,
"experimental": true,
"insecure-registries" : [],
"features": { "buildkit": true }
}
dest: /etc/docker/daemon.json
become: true
- name: enable docker service
service:
name: docker
enabled: yes
become: true
- name: add ubuntu user to docker group
user:
name: ubuntu
groups: docker
append: yes
become: true
- name: restart docker daemon
systemd:
state: restarted
daemon_reload: yes
name: docker
no_block: yes
become: true
# pull your images then run your containers

Answer 2

The only approach that I can think of is using AWS SSM to run multiple commands but still, you might need to spin 100's of instances and that would not be the good approach.
Below are the set of commands you can use :
Spin instance using below Cloudformation template, run it in loop to create multiple instances :
---
Resources:
MyInstance:
Type: AWS::EC2::Instance
Properties:
AvailabilityZone: <region>
ImageId: <amiID>
InstanceType: t2.micro
KeyName : <KeyName>
Use below command to get the intance ID :
aws ec2 describe-instances --filters 'Name=tag:Name,Values=EC2' --query 'Reservations[*].Instances[*].InstanceId' --output text
Using that instance-id, run below command :
aws ssm send-command --instance-ids "<instanceID>" --document-name "AWS-RunShellScript" --comment "<COMMENT>" --parameters commands='sudo yum update -y' --output text
I don't think docker will be of any help here as that would complicate things for you due to SSM agent installation. So your best bet would be running commands one by one and finally storing your output in S3.