Connecting to AWS S3

In this guide, we’ll link a private AWS S3 bucket to a Valohai project.

1. Requirements

For this tutorial you will need:

  • an Amazon Web Services (AWS) account you can administer
  • a Valohai project which to link the S3 bucket to

2. Create the S3 bucket

S3 home page

Create an S3 bucket through AWS console (https://s3.console.aws.amazon.com/s3/home).

2.1 Select bucket name and region

S3 bucket creation, page 1
  1. Throughout this guide, we will assume the name of the bucket is my-valohai-bucket; be sure to replace this with the actual name of your bucket when copying in any example configuration!
  2. Create the bucket in the region you’ll be running your training to minimize data transfer costs. If you don’t have a preference, we recommend using Ireland (eu-west-1) as most of our computation resides there.

2.2 Use default bucket properties

S3 bucket creation, page 2

Default bucket properties are fine, no need to activate versioning or anything else.

2.3 Use default bucket permissions

S3 bucket creation, page 3

Default bucket permissions are fine, but double check that your bucket is not public.

3. Configure CORS for the S3 bucket

If you wish to be able to upload files to the store using the app.valohai.com web UI, you will need to add a CORS policy document to the S3 bucket.

S3 bucket listing with the new bucket highlighted

First you navigate to the AWS S3 bucket you created.

S3 bucket CORS setting location

Then you go to the “CORS settings” and add the rules below.

<?xml version="1.0" encoding="UTF-8"?>
<CORSConfiguration xmlns="http://s3.amazonaws.com/doc/2006-03-01/">
    <CORSRule>
        <AllowedOrigin>*</AllowedOrigin>
        <AllowedMethod>GET</AllowedMethod>
        <MaxAgeSeconds>3000</MaxAgeSeconds>
        <AllowedHeader>Authorization</AllowedHeader>
    </CORSRule>
    <CORSRule>
        <AllowedOrigin>https://app.valohai.com</AllowedOrigin>
        <AllowedMethod>POST</AllowedMethod>
        <MaxAgeSeconds>3000</MaxAgeSeconds>
        <AllowedHeader>Authorization</AllowedHeader>
    </CORSRule>
</CORSConfiguration>

Now your bucket allows POSTs for your user on https://app.valohai.com website

4. Create an IAM user

IAM home page

Using the AWS console, start creating a new IAM user with programmatic access credentials (access key ID / secret access key).

4.1 Select name for your programmatic user

IAM user creation, page 1
  1. User name can be anything, try to be descriptive.
  2. Double check that programmatic access is turned on.

4.2 Skip the permission configuration

We will add permissions later, you can skip to the next step.

4.3 Save access key ID and secret for later usage

IAM user creation, page 3

Download the CSV or copy-paste the “Access key ID” and “Secret access key” somewhere safe.

Tip

If you lose these credentials, you can generate new ones though IAM > Select user > Security credentials > Create access key.

5. Allow the IAM user to access the bucket

Now we have a user without any permissions, let’s allow the user to access our new bucket.

IAM user listing

Find and open the user you created in the previous section.

IAM user inline policy location

Add a new inline policy. You can use any other AWS IAM policy definition methods just as well. Inline policies are the easiest get started.

IAM user inline policy definition

The user needs to have full access to the S3 bucket; an example of a suitable access policy document is below. Make sure to change the resource name my-valohai-bucket!

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": "s3:*",
            "Resource": [
                "arn:aws:s3:::my-valohai-bucket",
                "arn:aws:s3:::my-valohai-bucket/*"
            ]
        }
    ]
}
IAM user policy creation review page

Give your policy a descriptive name and we are done with the mandatory AWS setup!

5. Large file uploads (optional)

If executions need to upload outputs larger than 5 GB, additional setup is needed. This is optional and only required for large outputs.

To upload large outputs using Amazon’s multi-part upload API, a temporary AWS IAM role will be dispensed to the worker machines when required.

Be sure to replace the following placeholders in the following policy examples!

  • BUCKET – the target S3 bucket
  • ACCOUNTNUMBER – your AWS account number
  • USERNAME – the username liked to the access keys that are being used with the store

Create a new AWS IAM Role. The role policy document should look like:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "MultipartAccess",
            "Effect": "Allow",
            "Action": [
                "s3:AbortMultipartUpload",
                "s3:GetObject",
                "s3:ListBucket",
                "s3:ListBucketMultipartUploads",
                "s3:ListBucketVersions",
                "s3:ListMultipartUploadParts",
                "s3:PutObject"
            ],
            "Resource": [
                "arn:aws:s3:::BUCKET",
                "arn:aws:s3:::BUCKET/*"
            ]
        }
    ]
}

The trust relationship document should look like:

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Principal": {
        "AWS": "arn:aws:iam::ACCOUNTNUMBER:user/USERNAME"
      },
      "Action": "sts:AssumeRole"
    }
  ]
}

Take note of the role’s AWS ARN (arn:aws:...), that will be configured to your Valohai project.