Tempo

# AWS Setup

Note: These instructions assume the user has a moderate knowledge of AWS concepts, including S3 buckets, EC2 instances, AMIs, EBS volumes, AWS Batch, and autoscaling. For a succinct summary of these concepts, please check the AWS glossary, the AWS documentation (opens new window), and other recent AWS tutorials circa 2018.

Building the compute environment for AWS Batch consist of two steps.

NOTE

You need to first install (opens new window) and configure (opens new window) AWS command-line interface (CLI).

# Creating the S3 Bucket

You must first create an S3 bucket which will be used as working directory and where your outputs will be stored, as described here (opens new window).

Use S3 bucket path for <AWS-S3-WORKDIR> value when creating awsbatch.config.

# Building the AMI

Tempo's Amazon Machine Image (AMI) is implemented with EBS autoscaling script (opens new window) which will automatically increase disk space when needed.

To prepare the AMI for your compute environment, run the command below from the Tempo root directory.

aws cloudformation create-stack \
    --stack-name tempoAMI \
    --template-body file://aws_cf_scripts/AMICreate.yaml \
    --parameters ParameterKey=AMIName,ParameterValue=tempo-ami \
    --capabilities CAPABILITY_IAM

Note: You can specify any name instead of tempoAMI for stack-name.

This command submits CloudFormation Stack for building the AMI (opens new window) for execution, which we will use for building Batch Compute Environment. When you submit the command you will get StackId ARN as a response similar to this below.

{
    "StackId": "arn:aws:cloudformation:us-east-1:474622381158:stack/tempoAMI/f8d47a90-41c8-11e9-98cc-0eb85d5eff94"
}

Building the AMI will take a few minutes. You can check the status of the build process by running aws cloudformation describe-stacks --stack-name tempoAMI.

When build is complete the StackStatus field in the response JSON will have the value CREATE_COMPLETE and the Outputs section will look similar to:

"Outputs": [
                {
                    "OutputKey": "AMI",
                    "OutputValue": "ami-0599a84d7553f2fb7"
                }
            ]

We will use OutputValue for the next step.

# Building the Compute Environment

Run the command below from the Tempo repo root directory and set <AMI-ID> parameter to OutputValue from previous step. If you skipped previous section do not add the --parameters argument when running the command.

aws cloudformation create-stack \
    --stack-name tempoAWSBatchCE \
    --template-body file://aws_cf_scripts/AWSBatchCreate.yaml \
    --capabilities CAPABILITY_IAM \
    --parameters ParameterKey=AmiId,ParameterValue=<AMI-ID>

Note: You can specify any name instead of tempoAWSBatchCE for stack-name.

Building the AWS Batch Compute Environment will take a few minutes. You can check the status of the build process by running aws cloudformation describe-stacks --stack-name tempoAWSBatchCE.

When build is complete the StackStatus field in response JSON will have the value CREATE_COMPLETE and the Outputs section will look similar to:

"Outputs": [
                {
                    "OutputKey": "ComputeEnvironmentArn",
                    "OutputValue": "arn:aws:batch:us-east-1:474622381158:compute-environment/ComputeEnvironment-c5b2c9926a6fb42"conf/awsbatch.config
                },
                {
                    "OutputKey": "JobQueueArn",
                    "OutputValue": "arn:aws:batch:us-east-1:474622381158:job-queue/JobQueue-f2090879374b5cc"
                }
            ]

Use OutputValue of JobQueueArn for <AWS-BATCH-QUEUE-ARN> value when creating awsbatch.config.

# Create Configuration File

Create the awsbatch.config file in conf directory using conf/awsbatch.config.template

Replace the following values:

  • <AWS-REGION> with aws region where you built your compute environment.
  • <STORAGE-ENCRYPTION> storage encryption for your bucket (default: 'AES256').
  • <AWS-BATCH-QUEUE-ARN> ARN of your AWS Batch Job Queue built in Building the Compute Environment section.
  • <AWS-S3-WORKDIR> S3 bucket used as working directory created in Create the S3 Bucket section.

← Running the Pipeline →