Grace LogoGrace
Tutorials

Sequential Dataset → AWS S3

Learn how to automate migrating a z/OS sequential dataset to an AWS S3 bucket using Grace.

This tutorial will guide you through using Grace to orchestrate a simple but powerful hybrid workflow: moving a text-based sequential dataset from a z/OS mainframe to an AWS S3 bucket.

You will learn how to:

  • Define a Grace job that accesses a z/OS dataset.
  • Utilize Grace's shell job module to interact with the AWS CLI.
  • Manage character encoding to ensure your mainframe EBCDIC text data is correctly converted to a readable format (like UTF-8) in the cloud.
  • Automate the entire extraction and load process with a single grace run command.

This is a common scenario for archiving mainframe reports, making mainframe data available for cloud-based analytics, or staging data for further processing in cloud environments.

What you'll build

You'll write a grace.yml file that defines a single Grace job. This job will:

  1. Take a z/OS sequential dataset (containing EBCDIC text) as an input.
  2. Download this dataset to a local staging directory where Grace is running, converting it to your local text encoding (e.g., UTF-8).
  3. Upload this locally staged text file to an S3 bucket you specify.

Loaded sequential dataset TXT in the S3 bucket

Prerequisites

Before you begin, please ensure you have the following set up and configured:

  1. Grace CLI:
  2. Zowe CLI
    • Install Zowe CLI (v3 LTS recommended).
    • You must have a Zowe CLI zosmf profile that can successfully connect to your target z/OS system. This profile needs at least READ access to the sequential dataset you intend to migrate.
    • Verify your profile with zowe zosmf check status --your-profile-name.
    • For help, see Configuration.
  3. AWS CLI installed and configured:
    • Install the AWS CLI.
    • Configure it with AWS credentials that have permission to list buckets, create buckets (optional for this tutorial if you use an existing one), and upload objects (PutObject) to S3. Typically, this involves running aws configure and providing your Access Key ID, Secret Access Key, and default region.
    • This tutorial assumes usage of AWS CLI Version 2.
  4. Access to a z/OS system: Where your source sequential dataset resides.
  5. An AWS S3 bucket:
    • You'll need an S3 bucket where the file will be uploaded. If you don't have one, you'll create one in the first part of this tutorial.
  6. A text editor: For creating and editing your grace.yml file.

Prepare your mainframe data and S3 bucket

Before we define the Grace workflow, let's set up the necessary resources: a sample sequential dataset on the mainframe and an S3 bucket in AWS.

Create a sample sequential data set on z/OS

  1. Prepare the JCL

    Copy the following JCL and save it to a local file, for example, create_sample_seqfile.jcl. You can place this anywhere on your local machine; Grace itself won't directly use this file in the workflow, but you'll submit it using Zowe CLI.

    //CRSEQJOB JOB (ACCT),'SAMPLESEQ',CLASS=A,MSGCLASS=X
//**********************************************************************
//* THIS JOB CREATES A SAMPLE SEQUENTIAL DATASET WITH EBCDIC TEXT.
//*
//* !!! ACTION REQUIRED:
//*     REPLACE 'YOURHLQ' BELOW WITH YOUR TSO USER ID OR
//*     ANOTHER HIGH-LEVEL QUALIFIER YOU ARE AUTHORIZED TO USE.
//**********************************************************************
//STEP1    EXEC PGM=IEBGENER
//SYSPRINT DD SYSOUT=*
//SYSUT1   DD *,SYMBOLS=JCLONLY
THIS IS LINE 1 OF THE SAMPLE EBCDIC TEXT FILE.
GRACE ORCHESTRATES DATA MOVEMENT TO THE CLOUD.
LINE NUMBER 3 WITH SOME NUMBERS 12345 AND SYMBOLS: @#$
THE QUICK BROWN FOX JUMPS OVER THE LAZY DOG. 0123456789
/*
//SYSUT2   DD DSN=YOURHLQ.SAMPLE.TEXTFILE,
//            DISP=(NEW,CATLG,DELETE),
//            UNIT=SYSDA,
//            SPACE=(TRK,(1,1)),
//            DCB=(RECFM=FB,LRECL=80,BLKSIZE=8000)
//SYSIN    DD DUMMY

    ⚠️ Ensure you replace YOURHLQ in the SYSUT2 DD DSN statement with your actual TSO user ID or another High-Level Qualifier you are authorized to use.

  2. Upload and submit the JCL using Zowe CLI

    • First upload the JCL to a PDS on the mainframe (e.g. your personal JCL library). Replace YOUR.JCL.PDS(CRSEQJCL) with your actual PDS and desired member name.
    zowe zos-files upload file-to-data-set create_sample_seqfile.jcl "YOUR.JCL.PDS(CRSEQJCL)"
    • Then, submit the JCL:
    zowe zos-jobs submit data-set "YOUR.JCL.PDS(CRSEQJCL)" --view-all-spool-content
    • Verify that the job completes successfully (usually MAXCC 0000). If it fails, check the job output for errors (e.g. authority issues, invalid HLQ).

    You can view the created sequential dataset with:

    zowe zos-files view data-set "YOURHLQ.SAMPLE.TEXTFILE"

    Ensure the dataset identifier matches that of your JCL.

  3. Note your dataset name

    Once the job is successful, the dataset YOURHLQ.SAMPLE.TEXTFILE (with your actual HLQ) will exist on your z/OS system. Make a note of this full DSN, as you'll need it for your grace.yml file.

Create an S3 bucket

You'll need an S3 bucket to store the data extracted from the mainframe.

  1. Choose a globally unique bucket name and your desired AWS region.

  2. Open your terminal and use the AWS CLI:

    aws s3api create-bucket --bucket grace-demo-bucket --region [your_region]

    Make note of your exact bucket name, as you'll need it for grace.yml.

    ⚠️ S3 bucket names are globally unique. The grace-demo-bucket identifier you define will probably be something like grace-demo-bucket-[your_name]-[some_numbers].

Initialize your Grace project

Now, let's create a new Grace project directory.

  1. In your terminal, navigate to where you want to create your project and run:

    grace init my-seq-to-s3-flow

  2. Follow the interactive prompts:

    • Workflow name: You can accept the default (my-seq-to-s3-flow) or change it.
    • HLQ: Enter your typical High-Level Qualifier (e.g. your TSO ID). This will be used for datasets.jcl, datasets.src, etc. in grace.yml.
    • Profile: Enter the name of your working Zowe CLI profile.

  3. Enter your new project directory:

    cd my-seq-to-s3-flow

Configuring grace.yml

Open the grace.yml file that grace init created in your my-seq-to-s3-flow directory. We'll modify the config and datasets sections, and then define our single shell job.

A basic grace.yml will look something like this after grace init:

config:
  profile: your_zowe_profile # Populated by 'grace init'
 
datasets:
  jcl: YOURHLQ.WORKFLOW.JCL # Populated by 'grace init' with your HLQ
  src: YOURHLQ.WORKFLOW.SRC # Populated by 'grace init'
  loadlib: YOURHLQ.LOAD # Populated by 'grace init'
 
jobs:
  # We will add our job definition here

Verify/update the config block

  1. profile

    The grace init command should have populated this with the Zowe profile you entered. Ensure it's correct.

    config:
  profile: your_zowe_profile # Ensure this is correct

  2. defaults and concurrency

    For this specific tutorial, the defaults (for compiler/linker) and concurrency settings are not used because we are only defining a single shell job that doesn't compile or link code, nor does it content for mainframe job slots managed by Grace's concurrency setting.

    grace init does not generate these fields by default, so no further work is needed here.

Verify/update the datasets block

grace init should have populated the jcl, src, and loadlib fields with the HLQ you provided.

datasets:
  jcl: YOURHLQ.WORKFLOW.JCL
  src: YOURHLQ.WORKFLOW.SRC
  loadlib: YOURHLQ.LOAD

Ensure these DSNs look reasonable for your environment. Since we are not using grace deck to upload JCL or source for this tutorial (it's a shell job), these specific datasets are not critical to the workflow's function but are good to have correctly defined as a best practice.

grace init sanitizes the workflow name part of the DSN to conform to PDS member/qualifier rules (e.g. uppercasing, truncation, replacing hyphens).

Save your grace.yml file after ensuring everything looks good. In the next step, we'll write the jobs block.

Defining the shell job

Now, we will define the single job that orchestrates the data delivery. Add the following jobs block to your grace/yml file under the datasets block.

jobs:
  - name: SEQTOS3
    type: shell
    shell: bash
    inputs:
      - name: INFILE
        path: "zos://YOURHLQ.SAMPLE.TEXTFILE"
        encoding: text
    with:
      inline: |
        S3_BUCKET="your-s3-bucket-name"
        S3_KEY="migrated_data/seq_extract_$(date +%Y%m%d_%H%M%S).txt"
 
        aws s3 cp "$GRACE_INPUT_INFILE" "s3://$S3_BUCKET/$S3_KEY" --metadata Content-Type=text/plain
 
        if [ $? -eq 0 ]; then
          echo "Successfully uploaded to S3: s3://$S3_BUCKET/$S3_KEY"
        else
          exit 1 # Crucial: exit with non-zero to mark the Grace job as FAILED
        fi

⚠️ Change S3_BUCKET="your-s3-bucket-name" to point to the name of your S3 bucket.

⚠️ Change path: "zos://YOURHLQ.SAMPLE.TEXTFILE" to point to the DSN of your sequential dataset on the mainframe.

Key elements in this job:

  1. name: SEQTOS3: A clear name for this job

  2. type: shell: Specifies that this job will run local shell commands.

  3. shell: bash: Specifies that this job should use bash as its shell interpreter

  4. inputs:

    • name: INFILE: This logical name will be exposed as $GRACE_INPUT_INFILE inside the shell script.
    • path: "zos://YOURHLQ.SAMPLE.TEXTFILE":
      • The zos:// prefix tells Grace to wire this dataset to the shell job from the mainframe at runtime.
    • encoding: text: This is critical. It instructs Grace to download the EBCDIC file from the mainframe and convert it to the host's default text encoding (typically UTF-8 or another ACII-compatible encoding), making it readable by standard local tools like aws s3 cp.

  5. with:

    • inline: |

      We use an inline script for simplicity in this tutorial. The | operator enables multiline script definition.

      • S3_BUCKET="your-s3-bucket-name" Defines the target S3 bucket to push the dataset from the mainframe.

      • S3_KEY Defines the object key (filename and path within the bucket) for the uploaded file in S3. Using a timestamp makes each upload unique.

      • aws s3 cp "$GRACE_INPUT_INFILE" "s3://$S3_BUCKET/$S3_key" --metadata Content-type=text/plain

        • The core AWS CLI command.
        • It uses the $GRACE_INPUT_INFILE environment variable exposed by Grace, resolving to the input dataset we defined.
        • --metadata Content-Type=text/plain tells S3 to treat the uploaded object as a plain text file.

      • if [ $? -eq 0 ] ... else ... exit 1; fi

        Basic error checking for the AWS CLI command. Crucially, exit 1 on failure tells Grace that this shell job step has failed.

For details on virtual path prefixes like zos:// and how Grace exposes environment variables for job inputs and outputs, see Virtual Paths & Job I/O.

To learn more about how to work with the shell job module, see Jobs - Shell

After adding this jobs block, your complete grace.yml should look something like this:

config:
  profile: "your_zowe_profile"
 
datasets:
  jcl: "USER01.MYSEQ2S3.JCL"
  src: "USER01.MYSEQ2S3.SRC"
  loadlib: "USER01.LOAD"
 
jobs:
  - name: SEQTOS3
    type: shell
    inputs:
      - name: INFILE
        path: "zos://USER01.SAMPLE.TEXTFILE"
        encoding: text
    with:
      inline: |
        S3_BUCKET="grace-demo-bucket"
        S3_KEY="migrated_data/seq_extract_$(date +%Y%m%d_%H%M%S).txt"
 
        aws s3 cp "$GRACE_INPUT_INFILE" "s3://$S3_BUCKET/$S3_KEY" --metadata Content-Type=text/plain
 
        if [ $? -eq 0 ]; then
          echo "Successfully uploaded to S3: s3://$S3_BUCKET/$S3_KEY"
        else
          exit 1
        fi

Save your grace.yml file. Your workflow is now fully defined!

Running the Grace workflow

With your grace.yml file configured, you're now ready to execute the workflow. For this particular workflow, since we are not generating any JCL specific to Grace's z/OS job modules (like compile or linkedit) and are not uploading local source files via src:// prefixes for z/OS jobs, the grace deck command is not necessary. Grace will wire the z/OS dataset at runtime.

  1. Ensure you are in your project directory (e.g. my-seq-to-s3-flow) in your terminal session.

  2. Execute the workflow

    grace run

    You can also add the -v or --verbose flag for more detailed output, which is helpful for troubleshooting:

    grace run -v

As Grace runs the SEQTOS3 job, note key steps in the output:

  • Grace launching the SEQTOS3 job (🚀).
  • The final success messages (✅) from Grace.

If the aws s3 cp command fails (e.g. due to incorrect bucket name, permissions, or AWS CLI configuration), your script's exit 1 will cause Grace to mark the job as FAILED, and you'll see error messages.

Verifying the outcome

Loaded sequential dataset TXT in the S3 bucket

Once grace run completes successfully:

  1. Check AWS S3:

    Navigate to your S3 bucket in the AWS console or use the AWS CLI:

    aws s3 ls s3://grace-demo-bucket/migrated_data/

    You should see your uploaded file, e.g. seq_extract_YYYYMMDD_HHMMSS.txt.

  2. Inspect the uploaded file:

    aws s3 cp s3://grace-demo-bucket/migrated_data/seq_extract_YYYYMMDD_HHMMSS.txt ./downloaded_from_s3.txt
cat ./downloaded_from_s3.txt

    Verify that the content matches the sample data you created on the mainframe and that it's readable (i.e. correctly converted from EBCDIC to your local system's text encoding). It should look like this:

    THIS IS LINE 1 OF THE SAMPLE EBCDIC TEXT FILE.
GRACE ORCHESTRATES DATA MOVEMENT TO THE CLOUD.
LINE NUMBER 3 WITH SOME NUMBERS 12345 AND SYMBOLS: @#$
THE QUICK BROWN FOX JUMPS OVER THE LAZY DOG. 0123456789

  3. Review Grace logs (optional):

    • Navigate to the .grace/logs/ directory in your project.
    • Open the latest run-specific subdirectory (e.g. YYYYMMDDTHHMMSS_run_<workflow_uuid>).
    • You'll find:
      • SEQTOS3_shell-xxxxxx.json: The detailed JSON log for your shell job, containing the stdout and stderr from your inline script.
      • summary.json: The overall workflow summary.

Conclusion

Congratulations! You've successfully used Grace to automate the migration of a sequential text dataset from a z/OS mainframe to an AWS S3 bucket.

In this tutorial, you:

  • Defined a shell job in grace.yml.
  • Used inputs with a zos:// path and encoding: text to seamlessly wire mainframe data to a local script.
  • Leveraged environment variables ($GRACE_INPUT_*) within a shell script to work with data managed by Grace.
  • Automated against the mainframe and AWS cloud in a centralized control plane.

This simple example demonstrates a powerful pattern: using Grace to bridge mainframe systems with modern cloud services and local tooling.

Next steps to explore

  • Modify the shell job to perform transformations on the data locally before uploading to S3.
  • Create a workflow that first uses a z/OS job (e.g. running a COBOL program or DFSORT) to prepare or transform data on the mainframe into a zos-temp:// dataset, and then have a shell job download and upload that result.
  • Explore other Virtual Path Prefixes like src://, file://, zos-temp://, local-temp:// to learn about working with data across platforms and jobs.
  • Read the YAML Specification to discover more Grace features.

Hello, Grace!

Cover the basics of the Grace CLI and get up and running with your first workflow.

VSAM → PostgreSQL

Orchestrate a full ETL pipeline, inserting z/OS VSAM KSDS records into an AWS RDS Postgres instance using Grace.

On this page