How Watched Folder works

The Watched Folder service handles the creation, update, and deletion of the endpoints. After the administrator creates the endpoints, they are scheduled to be triggered by the Scheduler service based on the specified repeat interval or cron expression.

This diagram illustrates how Watched Folder processes an invocation request.

View full size graphic

The process of invoking a service using watched folders is as follows:

1. A client application places files or folders in the watched folder input folder.

2. When the job scan interval occurs, the Scheduler service invokes the provider.file_scan_service to process the files or folders in the input folder.

3. The provider.file_scan_service performs these tasks:

   • Scans the input folder for files or folders that match the include file pattern and excludes files or folders for the specified exclude file pattern. The oldest files or folders are picked up first. Files and folders that are older than the wait time are also picked up. In one scan, the number of files or folders that are processed are based on the batch size. For information about file patterns, see About file patterns. For information about setting the batch size, see Watched Folder service settings.

   • Picks up the files or folders for processing. If the files or folders are not completely downloaded, they are picked up in the next scan. To make sure that folders are completely downloaded, administrators should create a folder with a name by using the exclude file pattern. After the folder has all the files, it must be renamed to the pattern specified in the include file pattern. This step ensures that the folder has all the necessary files needed for invoking the service. For more information about ensuring that folders are completely downloaded, see Tips and tricks for watched folders.

   • Moves the files or folders to the stage folder after selecting them for processing.

   • Converts the files or folders in the stage folder to the appropriate input based on the endpoint input parameter mappings. For examples of input parameter mappings, see Tips and tricks for watched folders.

4. The target service configured for the endpoint is invoked either synchronously or asynchronously. The target service is invoked using the user name and password configured for the endpoint.

   • Synchronous invocation calls the target service directly and immediately handles the response.

   • For asynchronous invocation, the target service is called through the Job Manager service, which places the request in a queue. The Job Manager Service, in turn, calls the provider.file_write_results_service to handle the results.

5. The provider.file_write_results_service handles the response or failure of the target service invocation. When successful, the output is saved to the result folder based on the endpoint configuration. The provider.file_write_results_service also preserves the source if the endpoint is configured to preserve the results upon successful completion.

   When the invocation of the target service results in a failure, the provider.file_write_results_service logs the reason for the failure in a failure.log file and places that file in the failure folder. The failure folder is created based on the configuration parameters specified for the endpoint. When the administrator sets the Preserve On Failure option for the endpoint configuration, the provider.file_write_results_service also copies the source files into the failure folder. For information about recovering files from the failure folder, see Failure points and recovery.