Wait for Pipelines (deprecated)
Supported pipeline types:
|
The Wait for Pipelines processor is an orchestration stage that you use in orchestration pipelines. Orchestration stages perform tasks, such as schedule and start pipelines and Control Hub jobs, that you can use to create an orchestrated workflow across the StreamSets platform.
Use this processor when you want to wait for pipelines that were started upstream to complete, before performing other orchestration tasks. The Wait for Pipelines processor can wait for pipelines that are running on the specified Data Collector, Data Collector Edge, or Transformer execution engine. To wait for pipelines on multiple engines, use additional processors.
For example, you might use a Wait for Pipelines processor to wait for SDC Edge pipelines to complete that were started by a Start Pipelines origin. If you also wanted to wait for Data Collector pipelines to complete, you would use another Wait for Pipelines processor to perform that task.
The Wait for Pipelines processor checks the status of all pipelines listed in incoming orchestration records that run on the specified execution engine. When the pipelines complete, the processor updates the pipeline status details in the record and passes a single orchestration record downstream.
When you configure the Wait for Pipelines processor, you define the URL of the execution engine that runs the pipelines. For an execution engine registered with Control Hub, you specify the Control Hub URL, so the processor starts the pipelines through Control Hub. Then, you specify how long to wait between each pipeline status checks.
You also configure the user name and password to run the pipeline and can optionally configure SSL/TLS properties.
Stage Processing and Pipeline Implementation
Use the Wait for Pipelines processor downstream from a Start Pipelines origin or Start Pipelines processor that starts pipelines that run in the background. When running pipelines in the background, a Start Pipelines stage passes its orchestration record downstream immediately after starting pipelines, rather than waiting for them to complete.
When a Wait for Pipelines processor receives an orchestration record, it checks the status for all of the pipelines that are running on the execution engine specified in the stage. After all of the pipelines complete, the processor updates the pipeline status information in the orchestration record and passes the record downstream.
If you pass orchestration records from multiple stages to the processor, the processor waits until all pipelines associated with those records are complete, then passes a single merged orchestration record downstream.
For example, instead of using a Wait for Pipelines processor immediately after a Start Pipelines origin that starts a pipeline in the background, you can just configure the origin to run the pipeline in the foreground. Then, the Start Pipelines origin passes its orchestration record downstream after the pipeline completes, with no need for a Wait for Pipelines processor.
In contrast, say you want to start several SDC Edge and Transformer pipelines when you start your orchestration pipeline. You also want them all to complete before starting a Data Collector pipeline. To do this, you create the following pipeline:
You configure a Start Pipelines origin to start the SDC Edge pipelines in the background, which passes an orchestration record to the first Start Pipelines processor as soon as the jobs start. This enables the pipelines started by both stages to run concurrently.
You configure the Start Pipelines processor to run the Transformer pipelines in the foreground, so the processor passes the updated orchestration record downstream only after all of the pipelines complete. That takes care of the Transformer pipelines, but the SDC Edge pipelines may still be running.
To ensure that the SDC Edge pipelines complete before starting the Data Collector pipeline, you add the Wait for Pipelines processor and configure it to wait pipelines running on that Data Collector engine.
When the Wait for Pipelines processor receives the orchestration record, it notes the IDs of all of the pipelines started upstream that are running on the specified Data Collector, and waits for them to complete. After the pipelines complete, the Wait for Pipelines processor updates pipeline status information in the orchestration record and passes the record downstream, triggering the start of the Data Collector pipeline.
Generated Record
When the Wait for Pipelines processor completes its task, it updates the pipeline status and related information in the orchestration record before passing the record downstream.
Field Name | Description |
---|---|
<pipeline ID>/pipelineStatus | Status of the pipeline. For more
information, see Understanding Pipeline States. The processor updates this field. |
<pipeline ID>/pipelineStatusMessage | Status message for the
pipeline. The processor updates this field as needed. |
<pipeline ID>/finishedSuccessfully | Boolean field that
indicates whether a pipeline completed successfully. The processor adds this field. |
<pipeline ID>/pipelineMetrics | Map field that contains metrics for
the pipeline and individual pipeline stages. The processor adds these fields. |
<unique task name>/success | Boolean field that indicates whether all jobs
completed successfully.
The processor adds this field. |
For example, the following preview shows the fields that a Wait for Pipelines processor adds and updates in comparison to the incoming record:
Notice how the processor updated the pipelineStatus
field, and added the
finishedSuccessfully
, pipelineMetrics
, and
success
fields. All of the changes indicate that the pipeline
completed successfully.