s3store

• s3store
• Supported tools
  • Upload an entire folder to an S3 bucket
  • Download an entire bucket to a local folder
• Using Noobaa

s3store

Golang library to transfer files from and to S3 compatible storage.

• Targetted to run on K8s/OpenShift environments, can run as CLI application
• Easy to integrate as imported package in any client application

Supported tools

The following tools leverage the s3 Golang package to implement utility functions that simplify the client applications.

The tools are connected to the S3 service using the following environment variables:

• AWS_ACCESS_KEY_ID: the AWS access key
• AWS_SECRET_ACCESS_KEY: the AWS secret key
• S3_ENDPOINT: the S3 endpoint
• S3_REGION: the S3 region

This function in the s3filemanager package creates as AWS Session using the values from the above environment variables:

func ConnectWithEnvVariables() (*session.Session, error) {
  ...
}

Once the Session has been established, it is used to execute the tools.

Upload an entire folder to an S3 bucket

S3FolderUploader is a Golang struct that implements the function to upload a given local folder to a remote S3 bucket.

The bucket is created, if it doesn't exist.

All subfolders will be included, and the target bucket preserves the folder structure.

The following snippet creates and execute the S3FolderUploader tool:

  ...
  session, err := s3filemanager.ConnectWithEnvVariables()
  if err != nil {
    log.Fatal(err)
  }

  uploader := s3filemanager.NewS3FolderUploader(bucket, folder, logger)
  err = uploader.Run(session)
  ...

WithMatcher function allows to select only the files whose path contains the given matcher string.

WithPrexif function allows to add a fixed prefix to all the file names.

S3FolderUploader uses s3manager.Uploader to upload the entire folder recursively, using the UploadWithIterator scanner function to iterate through the files to be uploaded.

This implementation minimizes the memory footprint and the number of open files: files are read and closed just after the read is complete. It defines a lazyFileReader implementation of the io.Reader interface to postpone the actual file reading until it's really needed.

S3FolderUploader can be tested using the provided main function with the following options:

go run main.go -b BUCKET_NAME -debug -f SOURCE_FOLDER -m upload

Download an entire bucket to a local folder

S3BucketDownloader is a Golang struct that implements the function to fully download a remote S3 bucket to a local folder.

The destination folder is created, if it doesn't exist.

All subfolders will be included, and the target folder preserves the bucket structure.

The following snippet creates and execute the S3BucketDownloader tool:

  ...
  session, err := s3filemanager.ConnectWithEnvVariables()
  if err != nil {
    log.Fatal(err)
  }

  downloader := s3filemanager.NewS3BucketDownloader(bucket, folder, logger)
  err = downloader.Run(session)
  ...

S3BucketDownloader uses s3manager.Downloader to download the entire bucket, using the DownloadWithIterator function to iterate through the files to be downloaded.

This implementation minimizes the memory footprint and the number of open files: the downloaded content is bufferized by the bufferizedFileWriter implementation of the io.WriterAt interface, and the file is created and closed just after the download of the S3 object completes, by the SaveFileAfterWrite struct that extends s3manager.BatchDownloadObject.

S3BucketDownloader can be tested using the provided main function with the following options:

go run main.go -b BUCKET_NAME -debug -f DESTINATION_FOLDER -m download

Using Noobaa

These tools work with all S3-compatible endpoints, and Noobaa is a data gateway for objects that provides the same S3 API and management tools. If you can access a K8s/OpenShift cluster, you don't need an AWS account to create the S3-compatible storage.

Once you install the noobaa CLI, you can install the NooBaa operator and services on a given namespace as:

# Initialize the TARGET_NS
TARGET_NS= ...
noobaa install -n $TARGET_NS

You can get quick-start information by running:

oc descrive noobaa noobaa -n $TARGET_NS

and look for the section starting with:

  ...
  Readme:               

  Welcome to NooBaa!
  -----------------
  ...

To test the s3-tools functionality, initialize the S3 connection to the NooBaa instance first:

. ./noobaa-init.sh $TARGET_NS

The above script initializes all the environment variables for you.

Note: unless you manually specified a value in the field spec.region, the default S3 region for NooBaa instances is us-east-1

You can use the following commands to start playing with the S3 CLI (install aws CLI first):

# Create s3 alias
alias s3='aws --endpoint "https://$S3_ENDPOINT" --no-verify-ssl s3'

# list objects buckets
s3 ls
# list objects of bucket, recursively
s3 ls --recursive s3://mybucket
# count objects of bucket, recursively
s3 ls --summarize --recursive s3://mybucket

# remove objects from bucket, recursively
s3 rm --recursive s3://mybucket
# remove empty bucket
s3 rb s3://mybucket
# remove bucket and all content
s3 rb s3://mybucket --force