Provisioning Reference Guide › SPML Service › SPML Service Configuration › Retrying SPML Requests › Retry Architecture

Retry Architecture

You can configure the SPML service to retry asynchronous requests that act on objects residing on Endpoint Systems which cannot be contacted due to a transient failure. The flow of retries follows the standard asynchronous request processing in SPML.

• The client submits an asynchronous request to the SPML server with an additional operational attribute indicating that retries should be attempted on failure.
• If the request is processed without error, the request is marked as successfully processed.
• If the request's processing terminates with a hard failure, or its target object is not considered retriable by JIAM, an immediate failure results.
• If the request's processing terminates with a soft failure and the client has flagged the request as retriable, JIAM stores the request in the retry database and attempts to process it again after a configured interval. Prior to storage the current operation is simplified by removing any sub-operations which succeeded, so that only the sub-operations which failed due to soft failures will be retried.
• If a request has been retried a configured maximum number of times, then it is considered to have suffered a hard failure and JIAM discontinues the retries.

The following diagram gives an overview of the components of the retries:

Note that the SPML server is involved in the configuration of the retry persistence mechanism in JIAM but only JIAM actually adds and deletes records from it.