• An HTTP task is a request between  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.

• A HTTP task is defined as one of the tasks in a plan. For more information, see Plan View Page.

• This capability may need to be enabled in your environment. For more information, Overview of Operationalization.

Limitations

• Custom security certificates cannot be used.
• HTTP-based requests have a 30-second timeout limit.

Prerequisites

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.

Steps

1. Open your plan in Plan View. Click a node to create a new task.

2. In the right panel, select HTTP task.

3. Set the following parameters:

   Parameter	Description
   Name	User-visible name of the task.
   Url	URL where the webhook message is received by the other application.
   Headers	Insert HTTP content headers as key-value pairs. For example, if your body is in JSON format, you should include the following header: key: Content-Type value: application/json NOTE: You may be required to submit an authentication token as the value for the Authorization key. Please refer to the documentation for your receiving application about the required headers.
   Body	(POST, PUT, or PATCH methods only) The body of the request submitted to the receiving application. NOTE: If your request does not require a body, please insert {} here. This is a known issue.
   Method	Select the HTTP method to use to deliver the message. The appropriate method depends on the receiving application. Most use cases require the POST method.
   Secret key	(Optional) A secret key can be used to verify the webhook payload. This secret value must be inserted in this location, and it must be included as part of the code used to process the requests in the receiving application. Insert the secret value here as a string without quotes. For more information on how this secret key is used to generate a signature, see Verify Webhook Signatures below.
   Validate SSL certificate	When set to true, HTTPS (SSL) communications are verified to be using a valid certificate before transmission. NOTE: If you must send a request to an endpoint that has an expired/invalid certificate, you must disable SSL verification.
   Retry on failure	If the returned status code is outside of the 200-299 range, then the webhook is considered to have failed. When this option is enabled, the request is retried. When a message is retried, the following header is submitted: X-Http-Task-Guid

4. To test the connection, click Test. A success message is displayed.
5. To add the task to the flow, click Save.

Examples

Run another job

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

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

Prerequisites

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

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

key: Content-Type
value: application/json

key: Authorization
value: Bearer <paste your access token here>

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

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

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

key: Content-Type
value: application/json

{"text":"Your job has completed."}

Verify

1. Click Test to validate that this task will work. 
2. Run a job:
   1. 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. Examples:
  
  

  {{$http_xx.name}} {{$flow_xy.name}}

Syntax

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

{{$plan

{{$flow_

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

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

Optionally, you can configure the platform to sign the HTTP requests sent for a flow. 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:

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 Requests

Timestamp

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.