Skip to main content

Create HTTP Task

During the execution of your plan, you can create a task to send HTTP requests to a third-party application endpoint. For example, when a preceding task successfully executes, you can send an HTTP message to a designated endpoint with information from that task.

  • An HTTP task is a request between the Designer Cloud Powered by Trifacta platform and another application. These requests are delivered using over HTTP and can be interpreted by the receiving application to take action.

    Note

    Your receiving application may require that you whitelist the host and port number or IP address of the platform. Please refer to the documentation for your application.

  • An HTTP task is one of the task types available in a plan. For more information, see Plan View Page.

Limitations

  • Custom security certificates cannot be used.

  • HTTP-based requests have a 30-second timeout limit.

Prerequisites

Note

It's possible that webhook requests can be submitted back to the platform to execute API tasks within the platform. However, there are security concerns. Additional configuration is required. For more information, see Configure Webhooks.

Requirements for receiving application

To send an HTTP request to a target application, the application must be configured to receive the request:

  • Requests from outside of the application domain must be enabled.

    Note

    Your receiving application may require that you whitelist the host and port number or IP address of the platform. Please refer to the documentation for your application.

  • You must acquire the URL of the endpoint to which to send the HTTP request.

  • You must acquire any HTTP headers that must be inserted with each HTTP request.

  • If the request must be signed, additional configuration is required. Details are below.

Create Task

  1. Open your plan in Plan View. In your sequence of tasks. Click a Plus sign icon to create a new task.

  2. In the right panel, select HTTP task. The HTTP task panel is displayed.

    PlanViewPage-ViewForHTTPTask.png

    Figure: HTTP task

  3. Set the required parameters. For more information on parameters, see Plan View for HTTP Tasks.

  4. You can specify plan metadata information in the header values and request body of your request. For more information, see Plan Metadata References.

  5. To test the connection, click Test. A success message is displayed.

    Tip

    A status code of 200 indicates that the test was successful.

    Tip

    You can use the GET method for testing purposes. A GET request does not change any data on the target platform but may permit you to specify elements in the request body.

  6. To add the task, click Save.

Rename Task

To rename the task, click More menu > Edit in the right panel.

Tip

Good naming may include the target platform endpoint and method, as well as the purposes of the task in your plan.

Delete Task

To delete the task, click More menu > Delete. Confirm that you wish to delete the task.

Warning

This step cannot be undone.

Plan metadata references

Within the message of your other tasks, you can reference metadata about the plan, including this task. For more information, see Plan Metadata References.

Examples

Run another job

You can create a task to run another job on the successful execution of this one.

Tip

Use this method to create conditional sequences of job executions.

As needed, you can specify task overrides as part of launching a job via API. For more information, see API Task - Run Job.

Prerequisites

Note

For this example, the platform must be whitelisted to receive requests from itself. Additional configuration is required. For more information, see Configure Webhooks.

You must acquire the recipe identifier for the next job to execute.

  1. Open the flow containing the next recipe.

  2. In Flow View, click the recipe whose outputs you wish to generate.

  3. Review the URL for the recipe object. In the example below, the recipe Id value is 4:

    http://www.example.com:3005/flows/1?recipe=4&tab=recipe
  4. Retain this value for below.

Define the HTTP task

Parameter

Description

Name

This name appears in the Trifacta Application only.

Url

Specify the URL as follows, replacing the example values with your own:

http://www.example.com:3005/v4/jobGroups/

Headers

Insert the following two headers:

key: Content-Type
value: application/json
key: Authorization
value: Bearer <paste your access token here>

Note

The token value must be preceded by the string: Bearer.

Body

In the body, insert the recipe Id for the value for wrangledDataset, which is the internal platform term for recipe:

{
  "wrangledDataset": {
    "id": 4
  }
}

Method

Select the POST method.

Verify
  1. Run the plan for which the HTTP task was created.

  2. When the plan successfully completes, open the flow containing the other job to execute.

  3. When you select the target recipe, a new job should be queued, in-progress, or completed.

Slack channel message

Tip

Slack tasks are now a supported product feature. For more information, see Create Slack Task.

You can create an HTTP task to deliver a text message to a Slack channel of your choice.

Prerequisites

Set up your Slack installation to receive HTTP messages:

  1. If needed, create a Slack channel to receive your messages.

  2. Create an app.

  3. Activate incoming HTTP messages for your app.

  4. Specify the channel to receive your incoming messages.

  5. Copy the URL for the incoming HTTP request from the cURL statement.

Define the HTTP task

Parameter

Description

Name

This name appears in the Trifacta Application only.

Method

Select the POST method.

Url

Paste the URL that you copied from Slack.

Headers

Copy the content headers from the Slack cURL command:

key: Content-Type
value: application/json

Body

{"text":"Your job has completed."}
Verify
  1. Click Test to validate that this task will work.

  2. Run a job and check the Slack channel for a message.

Plan metadata examples

You can reference metadata information from the plan definition and the current plan run as part of the request of your HTTP task.

Notes:

  • You can only insert metadata references for tasks that have already occurred in the plan run before the HTTP task begins.

  • Each task in the current run is referenced using a two-letter code. Example:

    {{$http_xx.name}}
    
Syntax

A plan metadata reference is constructed using the following syntax. In the appropriate textbox, enter one of the following values:

Tip

Start by typing $, which provides access to a menu tree of metadata references for each of the metadata reference types. The final syntax is noted above.

Plans:

Metadata information from the plan definition or the current plan run:

{{$plan

Flows:

Metadata information for the flow tasks executed in the current plan run.

{{$flow_

Flow task:

Metadata information for the outputs generated by the specific flow task.

{{$flow_7p.['My Output Name'].

In this example:

  • flow_7p is a reference to the specific flow task.

  • 'My Output Name'is the display name for the underlying output.

Plan information

The following request body contains references to the Plan name, plan run identifier, and the flow that was just executed:

{"text":"Plan: {{$plan.name}} 
RunId: {{$plan.runId}}
Flow: {{$flow_7p.name}}
Success."}
Plan run information

The following request body contains plan execution information using timestamps:

{"text":"Plan: {{$plan.name}} 
RunId: {{$plan.runId}}
- plan start: {{$plan.startTime}}
Running time: {{$plan.duration}}

Times:
- last task start: {{$flow_7p.startTime}}
- last task end: {{$flow_7p.endTime}}
"}
HTTP task information

You can reference information from an HTTP task that has already occurred:

{"text":"{{$http_qg.name}} returned {{$http_qg.statusCode}}."} 
Flow task information

The following request body references information from a flow task in the plan:

{"text":"{{$flow_7p.name}} execution:
Duration: {{$flow_7p.duration}}
Status: {{$flow_7p.status}}


For more information, see jobIds: {{$flow_7p.jobIds}}
"}
Flow information

The following request body references information from the underlying output for the above flow task:

{"text":"Flow reference information:
Name: {{$flow_7p['2013 POS'].name}}
Favorite column: {{$flow_7p['2013 POS'].columns.Store_Nbr.name}} 
Least favorite data source: {{$flow_7p['2013 POS'].sources['POS-r01.txt'].name}}
For more information, see jobIds: {{$flow_7p.jobIds}}
"}

Notes:

  • You can reference columns from the generated results using the .columns. reference.

  • You can reference information from datasources using the .sources reference.

For more information, see Plan Metadata References.

Feed metadata inputs to cloud function

This example demonstrates how you can use an HTTP task to deliver plan metadata to AWS lambda functions. A similar approach could be used for Google Cloud functions.

In this case, the rowCount value from the flow task execution is delivered via HTTP task to an AWS lambda function.

General steps:

  1. Define your plan.

  2. Flow task: Run the flow to generate the outputs needed for your Lamda function.

  3. HTTP task: generates an HTTP request whose body includes a reference to the rowCount metadata variable. Request body:

    {
     "rowCount": "{{$flow_7p['My Flow Name'].output['My output name'].rowCount}}"
    }
  4. AWS Lambda functions: The following is pseudo-code for Lambda:

    import json
    def lambda_handler(event, context):
      httpTaskBody = json.loads(event["body"])
      rowCount = httpTaskBody["rowCount"]
    
      return {
        'statusCode': 200,
        'body': json.dumps(rowCount)
      }
  5. Google Cloud functions: The following is pseudo-code for Google Cloud functions:

    def get_row_count(request):
      request_json = request.get_json()
      if request_json and 'rowCount' in request_json:
            rowCount = request_json['rowCount']
        return rowCount
      return 'No rowCount attribute provided'

Verify Signatures

Warning

Depending on the target application, implementing signature verification may require developer skills.

Optionally, you can configure the platform to sign the HTTP requests. Signed requests guarantee that the requests are sent from the platform, instead of a third party.

Below, you can review how the signature is created, so that you can configure the receiving application to properly process the signature and its related request.

Signature Header

HTTP requests are signed by inserting the X-Webhook-Signature header in the request. These signatures are in the following form:

X-Webhook-Signature: t=<timestamp>,sha256=<signature>

where:

  • <timestamp> - Timestamp when the signature was sent. Value is in UNIX time.

  • <signature> - SHA256 signature. The platform generates this signature using a hash-based message authentication code (HMAC) with SHA-256.

More information on these values is available below.

Example:

X-Webhook-Signature: t=1568818215724,sha256=55fa71b2e391cd3ccba8413fb51ad16984a38edb3cccfe81f381c4b8197ee07a
Check Application Tools

Depending on the application, you may need to complete one of the following sets of tasks to verify the task signatures:

Note

You may need to whitelist the platform in your application. See the application's documentation for details.

You may be required to create some custom coding for your application. Below, you can review details on how to do so, including a JavaScript example.

Process Signed RequestsTimestamp

The timestamp value (t=<timestamp>) appears at the beginning of the header value to prevent replay attacks, where an attacker could intercept a valid payload and its signature and re-transmit them.

  • To avoid such attacks, a timestamp is included in the signature header and is also embedded as part of the signed payload.

  • Since the timestamp is part of the signed payload, an attacker cannot change the timestamp value without invalidating the signature.

    • If the signature is valid but the timestamp is too old, you can then choose to reject the request.

    • For example, if you receive a request with a timestamp that corresponds to a date from one hour ago, you should probably reject the request.

  • For more information on replay attacks, see https://en.wikipedia.org/wiki/Replay_attack.

Signature

The task signature includes as part of its hashed value:

  • The secret key (entered above)

  • The timestamp value

  • Request data:

    • (POST/PUT/PATCH) - the body of the request

    • (GET/DELETE) - URL of the request

Step 1 - Extract the timestamp and signatures

Split the X-Webhook-Signature header:

  1. Split values using the , character as a separator.

  2. Split each of the parts using the = character.

  3. Extract the values for the timestamp and signature. From the above example:

    1. timestamp: 1568818215724

    2. signature: 55fa71b2e391cd3ccba8413fb51ad16984a38edb3cccfe81f381c4b8197ee07a

Step 2 - Create the expected signature

In the receiving application, you can recompute the signature to verify that the request was sent from the platform.

  1. Concatenate the timestamp, the dot character . and the request body (POST/PUT/PATCH methods) or the url (GET/DELETE methods).

  2. Suppose the above example is the signature for a POST request, and the request body is test. The concatenated value is the following:

    1568818215724.test
  3. You can now compute the HMAC authentication code in your receiving application. In the following JavaScript example, the secret key value is mySecret:

    const crypto = require('crypto');
    
    const message = '1568818215724.test'; // as defined above
    
    const hmac = crypto.createHmac('sha256', 'mySecret');
    hmac.update(message)
    const expectedSignature = hmac.digest('hex');
Step 3 - Compare the signatures

The value returned by your code and the value included as the signature in the X-Webhook-Signature header should be compared:

  • If the values do not match, reject the request.

  • If the values do match, compute the difference between the current timestamp and the timestamp in the header. If the difference is outside of your permitted limit, reject the request.

  • Otherwise, process the request normally in your application.