request_reshuffle #

Description #

request_reshuffle can analyze non-bulk requests to Elasticsearch, archive them in a queue, and store them on disk first. This allows business-side requests to return quickly, decoupling the front-end writes from the back-end Elasticsearch cluster. request_reshuffle requires offline pipeline consumption tasks to work in conjunction.

Configuration Example #

Here is a simple example:

flow:
  - name: backup-flow-request-reshuffle
    filter:
      - flow:
          flows:
            - set-auth-for-backup-flow
      - request_reshuffle: # Reshuffle none-bulk requests
          elasticsearch: "backup"
          queue_name_prefix: "request_reshuffle"
          partition_size: $[[env.REQUEST_RESHUFFLE_PARTITION_SIZE]]
          tag_on_success: [ "commit_message_allowed" ]

Parameter Description #

Name	Type	Description
elasticsearch	string	Elasticsearch cluster instance name
queue_name_prefix	string	Queue name prefix, default is `async_bulk`, default Label is `type:request_reshuffle`
partition_size	int	In addition to `level`, partitioning is based on the document `_id`. This parameter sets the maximum partition size.
continue_after_reshuffle	bool	Whether to continue with subsequent processes after Reshuffle is complete, default is `false`
tag_on_success	array	Add specified tags to the request context after processing all bulk requests