chainer-cfn: Cloudformation Template for ChainerMN on AWS
This template automates to build ChainerMN cluster on AWS. The overview of AWS resources to be created by this template are below:
- VPC and Subnet where cluster places (you can configure existing VPC/Subnet)
- S3 Bucket for sharing ephemeral ssh-key which is used to communicate among MPI processes in the cluster
- Placement group for optimizing network performance
- ChainerMN cluster which consists
1master EC2 instanceN (>=0)worker instnaces (via AutoScalingGroup)chaineruser to run mpi job in each instancehostfileto run mpi job in each instance- All the instances are launched from Chainer AMI
- (Option) Amazon Elastic Filesystem (you can configure existing filesystem)
- This is mounted on cluster instances automatically to share your code and data.
- Several required SecurityGroups, IAM Role
Please see template/main.py for detailed resource definitions.
The Latest Published Template
Quick Start
Please also refer to our blog: ChainerMN on AWS with CloudFormation
Development Manual
How to build a template
make build
How to test
# Configure AWS account properly first.
# this will create a stack via a template you built.
make create-stack TEST_STACK=YOUR_TEST_STACK_NAME KEY_PAIR_NAME=YOUR_KEY_PAIR_NAME
# perform ChainerMN's train_mnist.py
make e2e-test TEST_STACK=YOUR_TEST_STACK_NAME KEY_PAIR_NAME=YOUR_KEY_PAIR_NAME
# cleanup stack
make delete-stack TEST_STACK=YOUR_TEST_STACK_NAME KEY_PAIR_NAME=YOUR_KEY_PAIR_NAME
How to release
# Configure AWS account properly first.
# build template
make build
# perform e2e test
make create-stack TEST_STACK=YOUR_TEST_STACK_NAME KEY_PAIR_NAME=YOUR_KEY_PAIR_NAME
make e2e-test TEST_STACK=YOUR_TEST_STACK_NAME KEY_PAIR_NAME=YOUR_KEY_PAIR_NAME
make delete-stack TEST_STACK=YOUR_TEST_STACK_NAME KEY_PAIR_NAME=YOUR_KEY_PAIR_NAME
# publish to stage
make publish STAGE=(production|staging)
Release Notes
Version 0.1.0
- Initial release
- Based on Chainer AMI
0.1.0
- Based on Chainer AMI
- Released Template
License
MIT License (see LICENSE file).

Formed in 2009, the Archive Team (not to be confused with the archive.org Archive-It Team) is a rogue archivist collective dedicated to saving copies of rapidly dying or deleted websites for the sake of history and digital heritage. The group is 100% composed of volunteers and interested parties, and has expanded into a large amount of related projects for saving online and digital history.

