Pipeline #
What Is Pipeline? #
A pipeline is a function combination used for processing tasks offline. It uses the pipeline design pattern, just as online request filters do. A processor is the basic unit of a pipeline. Each processing component focuses on one task and the components can be flexibly assembled, and plugged and removed as required.
Pipeline Definition #
A typical pipeline service is defined as follows:
pipeline:
- name: request_logging_index
auto_start: true
keep_running: true
processor:
- json_indexing:
index_name: "gateway_requests"
elasticsearch: "dev"
input_queue: "request_logging"
idle_timeout_in_seconds: 1
worker_size: 1
bulk_size_in_mb: 10 #in MB
In the above configuration, a processing pipeline named request_logging_index is defined, and the processor parameter defines several processing units for the pipeline, which are executed in sequence.
Parameter Description #
Parameters related to pipeline definition are described as follows:
| Name | Type | Description |
|---|---|---|
| name | string | Name of a pipeline, which must be unique |
| auto_start | bool | Whether the pipeline automatically starts with the gateway startup, that is, whether the task is executed immediately |
| keep_running | bool | Whether the gateway starts executing the task again after completing the execution |
| singleton | bool | Whether the task is a singleton and only one node instance is allowed to run in a cluster |
| max_running_in_ms | int | The maximum time that the task runs execution, 60000 milliseconds by default. |
| retry_delay_in_ms | int | Minimum waiting time for the task re-execution, which is set to 5000 milliseconds by default |
| processor | array | List of processors to be executed by the pipeline in sequence |