Page tree


Outdated release! Latest docs are Release 8.7: Enable S3 Access

   

Contents:


Below are instructions on how to configure Trifacta® Self-Managed Enterprise Edition to point to S3.

  • Simple Storage Service (S3) is an online data storage service provided by Amazon, which provides low-latency access through web services. For more information, see https://aws.amazon.com/s3/.

 

NOTE: Please review the limitations before enabling. See 331262896 below.

Base Storage Layer

  • If base storage layer is S3: you can enable read/write access to S3.
  • If base storage layer is not S3: you can enable read-only access to S3.

Limitations

  • The Trifacta platform only supports running S3-enabled instances over AWS.
  • Access to AWS S3 Regional Endpoints through internet protocol is required. If the machine hosting the Trifacta platform is in a VPC with no internet access, a VPC endpoint enabled for S3 services is required. The Trifacta platform does not support access to S3 through a proxy server.
  • Write access requires using S3 as the base storage layer. See Set Base Storage Layer.

NOTE: Spark 2.3.0 jobs may fail on S3-based datasets due to a known incompatibility. For details, see https://github.com/apache/incubator-druid/issues/4456.

If you encounter this issue, please set spark.version to 2.1.0 in platform configuration. For more information, see Admin Settings Page.

Pre-requisites

  • On the Trifacta node, you must install the Oracle Java Runtime Environment for Java 1.8. Other versions of the JRE are not supported. For more information on the JRE, see http://www.oracle.com/technetwork/java/javase/downloads/index.html
  • If IAM instance role is used for S3 access, it must have access to resources at the bucket level.

Required AWS Account Permissions

For more information, see Required AWS Account Permissions.

Configuration

Depending on your S3 environment, you can define:

  • read access to S3
  • access to additional S3 buckets
  • S3 as base storage layer
  • Write access to S3
    • S3 bucket that is the default write destination

Define base storage layer

The base storage layer is the default platform for storing results.

Required for:

  • Write access to S3
  • Connectivity to Redshift

The base storage layer for your Trifacta instance is defined during initial installation and cannot be changed afterward.

If S3 is the base storage layer, you must also define the default storage bucket to use during initial installation, which cannot be changed at a later time. See Define default S3 write bucket below.

For more information on the various options for storage, see Storage Deployment Options.

For more information on setting the base storage layer, see Set Base Storage Layer.

Enable job output manifest

When the base storage layer is set to S3, you must enable the platform to generate job output manifest files.

During job execution, the platform creates a manifest file of all files generated during job execution. This manifest file is used as a reference to indicate the files that eventually appear, as there can be consistency issues between the files that S3 reports as present and the files that are actually written. The manifest assists the platform in determining when the files are actually ready for use.

  • Manifest filenames begin with an underscore.
  • Manifest files are small.
  • Manifest files may be written even for single-file outputs.

NOTE: Avoid deleting manifest files. After results have been generated, the manifest file may be used in the future for additional writing and publishing operations.


When the job results are published, this manifest file ensures proper publication.

NOTE: This feature must be enabled when using S3 as the base storage layer.

Steps:

  1. You can apply this change through the Admin Settings Page (recommended) or
    trifacta-conf.json
    . For more information, see Platform Configuration Methods.
  2. Locate the following parameter and set it to true:

    "feature.enableJobOutputManifest": true,
  3. Save your changes and restart the platform.

Enable read access to S3

When read access is enabled, Trifacta users can explore S3 buckets for creating datasets. 

NOTE: When read access is enabled, Trifacta users have automatic access to all buckets to which the specified S3 user has access. You may want to create a specific user account for S3 access.


NOTE: Data that is mirrored from one S3 bucket to another might inherit the permissions from the bucket where it is owned.

Steps:

  1. You apply this change through the Workspace Settings Page. For more information, see Platform Configuration Methods.
  2. Set the following property to enabled:

    Enable S3 Connectivity
  3. Save your changes.
  4. In the S3 configuration section, set enabled=true, which allows Trifacta users to browse S3 buckets through the Trifacta application.
  5. Specify the AWS key and secret values for the user to access S3 storage.

Configure file storage protocols and locations

The Trifacta platform must be provided the list of protocols and locations for accessing S3. 

Steps:

  1. You can apply this change through the Admin Settings Page (recommended) or
    trifacta-conf.json
    . For more information, see Platform Configuration Methods.
  2. Locate the following parameters and set their values according to the table below:

    "fileStorage.whitelist": ["s3"],
    "fileStorage.defaultBaseUris": ["s3:///"],
    ParameterDescription
    filestorage.whitelist

    A comma-separated list of protocols that are permitted to access S3.

    NOTE: The protocol identifier "s3" must be included in this list.

    filestorage.defaultBaseUris

    For each supported protocol, this parameter must contain a top-level path to the location where Trifacta platform files can be stored. These files include uploads, samples, and temporary storage used during job execution.

    NOTE: A separate base URI is required for each supported protocol. You may only have one base URI for each protocol.

    NOTE: For S3, three slashes at the end are required, as the third one is the end of the path value. This value is used as the base URI for all S3 connections created in Trifacta Self-Managed Enterprise Edition.

    Example:

    s3:///

    The above example is the most common example, as it is used as the base URI for all S3 connections that you create. Do not add a bucket value to the above URI.

  3. Save your changes and restart the platform.

Java VFS service

Use of SFTP connections requires the Java VFS service in the Trifacta platform.

NOTE: This service is enabled by default.

For more information on configuring this service, see Configure Java VFS Service.

S3 access modes

The Trifacta platform supports the following modes for access S3. You must choose one access mode and then complete the related configuration steps.

NOTE: Avoid switching between user mode and system mode, which can disable user access to data. At install mode, you should choose your preferred mode.

 

System mode

(default) Access to S3 buckets is enabled and defined for all users of the platform. All users use the same AWS access key, secret, and default bucket.

System mode - read-only access

For read-only access, the key, secret, and default bucket must be specified in configuration.

NOTE: Please verify that the AWS account has all required permissions to access the S3 buckets in use. The account must have the ListAllMyBuckets ACL among its permissions.

Steps:

  1. You can apply this change through the Admin Settings Page (recommended) or
    trifacta-conf.json
    . For more information, see Platform Configuration Methods.

  2. Locate the following parameters:

    ParametersDescription
    aws.s3.keySet this value to the AWS key to use to access S3.
    aws.s3.secretSet this value to the secret corresponding to the AWS key provided.
    aws.s3.bucket.name

    Set this value to the name of the S3 bucket from which users may read data.

    NOTE: Bucket names are not validated.

    NOTE: Additional buckets may be specified. See below.

  3. Save your changes.

User mode

Optionally, access to S3 can be defined on a per-user basis. This mode allows administrators to define access to specific buckets using various key/secret combinations as a means of controlling permissions.

NOTE: When this mode is enabled, individual users must have AWS configuration settings applied to their account, either by an administrator or by themselves. The global settings in this section do not apply in this mode.

To enable:

  1. You apply this change through the Workspace Settings Page. For more information, see Platform Configuration Methods.

  2. Verify that the following setting has been set to enabled:

    Enable S3 Connectivity
  3. You can apply this change through the Admin Settings Page (recommended) or

    trifacta-conf.json
    . For more information, see Platform Configuration Methods.

  4. Please verify that the setting below has been configured:

    "aws.mode": "user",
  5. Additional configuration is required for per-user authentication. 

    1. You can choose to enable session tags to leverage your existing S3 permission scheme.

    2. For more information, see Configure AWS Per-User Authentication.
User mode - Create encryption key file

If you have enabled user mode for S3 access, you must create and deploy an encryption key file. For more information, see Create Encryption Key File.

NOTE: If you have enabled user access mode, you can skip the following sections, which pertain to the system access mode, and jump to the Enable Redshift Connection section below.

System mode - additional configuration

The following sections apply only to system access mode.

Define default S3 write bucket

When S3 is defined as the base storage layer, write access to S3 is enabled. The Trifacta platform attempts to store outputs in the designated default S3 bucket. 

NOTE: This bucket must be set during initial installation. Modifying it at a later time is not recommended and can result in inaccessible data in the platform.

NOTE: Bucket names cannot have underscores in them. See http://docs.aws.amazon.com/AmazonS3/latest/dev/BucketRestrictions.html.

Steps:

  1. Define S3 to be the base storage layer. See Set Base Storage Layer.
  2. Enable read access. See Enable read access.
  3. Specify a value for  aws.s3.bucket.name  which defines the S3 bucket where data is written. Do not include a protocol identifier. For example, if your bucket address is s3://MyOutputBucket, the value to specify is the following:

    MyOutputBucket

    NOTE: Bucket names are not validated.

    NOTE: Specify the top-level bucket name only. There should not be any backslashes in your entry.

NOTE: This bucket also appears as a read-access bucket if the specified S3 user has access.

Adding additional S3 buckets

When read access is enabled, all S3 buckets of which the specified user is the owner appear in the Trifacta application. You can also add additional S3 buckets from which to read.

NOTE: Additional buckets are accessible only if the specified S3 user has read privileges.

NOTE: Bucket names cannot have underscores in them.

Steps:

  1. You can apply this change through the Admin Settings Page (recommended) or

    trifacta-conf.json
    . For more information, see Platform Configuration Methods.

  2. Locate the following parameter: aws.s3.extraBuckets:

    1. In the Admin Settings page, specify the extra buckets as a comma-separated string of additional S3 buckets that are available for storage. Do not put any quotes around the string. Whitespace between string values is ignored.

    2. In 

      trifacta-conf.json
      , specify the extraBuckets array as a comma-separated list of buckets as in the following: 

      "extraBuckets": ["MyExtraBucket01","MyExtraBucket02","MyExtraBucket03"]

      NOTE: Specify the top-level bucket name only. There should not be any backslashes in your entry.

      NOTE: Bucket names are not validated.

  3. Specify the extraBuckets array as a comma-separated list of buckets as in the following: 

    "extraBuckets": ["MyExtraBucket01","MyExtraBucket02","MyExtraBucket03"]
  4. These values are mapped to the following bucket addresses:

    s3://MyExtraBucket01
    s3://MyExtraBucket02
    s3://MyExtraBucket03

S3 Configuration

Configuration reference

You apply this change through the Workspace Settings Page. For more information, see Platform Configuration Methods.

SettingDescription
Enable S3 ConnectivityWhen set to enabled, the S3 file browser is displayed in the GUI for locating files. For more information, see S3 Browser.

You can apply this change through the Admin Settings Page (recommended) or

trifacta-conf.json
. For more information, see Platform Configuration Methods.

"aws.s3.bucket.name": "<BUCKET_FOR_OUTPUT_IF_WRITING_TO_S3>"
"aws.s3.key": "<AWS_KEY>",
"aws.s3.secret": "<AWS_SECRET>",
"aws.s3.extraBuckets": ["<ADDITIONAL_BUCKETS_TO_SHOW_IN_FILE_BROWSER>"]
SettingDescription
bucket.name 

Set this value to the name of the S3 bucket to which you are writing.

  • When webapp.storageProtocol is set to s3, the output is delivered to aws.s3.bucket.name.
key

Access Key ID for the AWS account to use.

NOTE: This value cannot contain a slash (/).

secret

Secret Access Key for the AWS account.

extraBuckets

Add references to any additional S3 buckets to this comma-separated array of values.

The S3 user must have read access to these buckets.

Enable use of server-side encryption

You can configure the Trifacta platform to publish data on S3 when a server-side encryption policy is enabled. SSE-S3 and SSE-KMS methods are supported. For more information, see http://docs.aws.amazon.com/AmazonS3/latest/dev/serv-side-encryption.html.

Notes:

  • When encryption is enabled, all buckets to which you are writing must share the same encryption policy. Read operations are unaffected.

To enable, please specify the following parameters.

You can apply this change through the Admin Settings Page (recommended) or

trifacta-conf.json
. For more information, see Platform Configuration Methods.

Server-side encryption method

"aws.s3.serverSideEncryption": "none",

Set this value to the method of encryption used by the S3 server. Supported values:

NOTE: Lower case values are required.

 

  • sse-s3
  • sse-kms
  • none

Server-side KMS key identifier

When KMS encryption is enabled, you must specify the AWS KMS key ID to use for the server-side encryption.

"aws.s3.serverSideKmsKeyId": "",

Notes:

The format for referencing this key is the following:

"arn:aws:kms:<regionId>:<acctId>:key/<keyId>"

You can use an AWS alias in the following formats. The format of the AWS-managed alias is the following:

"alias/aws/s3"

The format for a custom alias is the following:

"alias/<FSR>"

where:

<FSR> is the name of the alias for the entire key.

Save your changes and restart the platform.

Configure S3 filewriter

The following configuration can be applied through the Hadoop site-config.xml file. If your installation does not have a copy of this file, you can insert the properties listed in the steps below into

trifacta-conf.json
to configure the behavior of the S3 filewriter.

Steps:

  1. You can apply this change through the Admin Settings Page (recommended) or

    trifacta-conf.json
    . For more information, see Platform Configuration Methods.

  2. Locate the filewriter.hadoopConfig block, where you can insert the following Hadoop configuration properties:

    "filewriter": {
      max: 16,
      "hadoopConfig": {
        "fs.s3a.buffer.dir": "/tmp",
        "fs.s3a.fast.upload": "true"
      },
      ...
    }
    PropertyDescription
    fs.s3a.buffer.dir

    Specifies the temporary directory on the Trifacta node to use for buffering when uploading to S3. If fs.s3a.fast.upload is set to false, this parameter is unused.

    NOTE: This directory must be accessible to the Batch Job Runner process during job execution.

    fs.s3a.fast.upload

    Set to true to enable buffering in blocks.

    When set to false, buffering in blocks is disabled. For a given file, the entire object is buffered to the disk of the Trifacta node. Depending on the size and volume of your datasets, the node can run out of disk space.

  3. Save your changes and restart the platform.

Create Redshift Connection

For more information, see Create Redshift Connections.

Hadoop distribution-specific configuration

Hortonworks

NOTE: If you are using Spark profiling through Hortonworks HDP on data stored in S3, additional configuration is required. See Configure for Hortonworks.

Additional Configuration for S3

The following parameters can be configured through the Trifacta platform to affect the integration with S3. You may or may not need to modify these values for your deployment.

You can apply this change through the Admin Settings Page (recommended) or

trifacta-conf.json
. For more information, see Platform Configuration Methods.

ParameterDescription
aws.s3.consistencyTimeout

S3 does not guarantee that at any time the files that have been written to a directory will be consistent with the files available for reading. S3 does guarantee that eventually the files are in sync.

This guarantee is important for some platform jobs that write data to S3 and then immediately attempt to read from the written data.

This timeout defines how long the platform waits for this guarantee. If the timeout is exceeded, the job is failed. The default value is 120.

Depending on your environment, you may need to modify this value.

aws.s3.endpoint

This value should be the S3 endpoint DNS name value.

NOTE: Do not include the protocol identifier.

Example value:

s3.us-east-1.amazonaws.com

If your S3 deployment is either of the following:

  • located in a region that does not support the default endpoint, or
  • v4-only signature is enabled in the region

Then, you can specify this setting to point to the S3 endpoint for Java/Spark services.

For more information on this location, see https://docs.aws.amazon.com/general/latest/gr/rande.html#s3_region.

Testing

Restart services. See Start and Stop the Platform.

Try running a simple job from the Trifacta application. For more information, see Verify Operations.

Troubleshooting

Profiling consistently fails for S3 sources of data

If you are executing visual profiles of datasets sourced from S3, you may see an error similar to the following in the batch-job-runner.log file:

01:19:52.297 [Job 3] ERROR com.trifacta.hadoopdata.joblaunch.server.BatchFileWriterWorker - BatchFileWriterException: Batch File Writer unknown error:
{jobId=3, why=bound must be positive}
01:19:52.298 [Job 3] INFO com.trifacta.hadoopdata.joblaunch.server.BatchFileWriterWorker - Notifying monitor for job 3 with status code FAILURE

This issue is caused by improperly configuring buffering when writing to S3 jobs. The specified local buffer cannot be accessed as part of the batch job running process, and the job fails to write results to S3.

Solution:

You may do one  of the following:

  • Use a valid temp directory when buffering to S3.
  • Disable buffering to directory completely.

Steps:

  • You can apply this change through the Admin Settings Page (recommended) or

    trifacta-conf.json
    . For more information, see Platform Configuration Methods.

  • Locate the following, where you can insert either of the following Hadoop configuration properties:

    "filewriter": {
      max: 16,
      "hadoopConfig": {
        "fs.s3a.buffer.dir": "/tmp",
        "fs.s3a.fast.upload": false
      },
      ...
    }
    PropertyDescription
    fs.s3a.buffer.dir

    Specifies the temporary directory on the Trifacta node to use for buffering when uploading to S3. If fs.s3a.fast.upload is set to false, this parameter is unused.

    fs.s3a.fast.upload
    When set to false, buffering is disabled.
  • Save your changes and restart the platform.

Spark local directory has no space

During execution of a Spark job, you may encounter the following error:

org.apache.hadoop.util.DiskChecker$DiskErrorException: No space available in any of the local directories.

Solution:

  • Restart Trifacta services, which may free up some temporary space.
  • Use the steps in the preceding solution to reassign a temporary directory for Spark to use (fs.s3a.buffer.dir). 

 

This page has no comments.