It is possible for a request to complete prematurely multiple times. Each time, you must restart the request until it completes normally. The following series of events shows how to handle these requests:

1. The request completes prematurely and the system returns a restart token in the answer area.
2. Process any information that may have been returned from the request.
3. Issue IXLCACHE to restart the request. You must code RESTOKEN or EXTRESTOKEN to specify the restart token. All other keywords coded on the original request need to be coded on the restart request.
4. The request again completes prematurely and the system returns a restart token in the answer area.
5. Process any information that may have been returned from the request.
6. Issue IXLCACHE with the RESTOKEN or EXTRESTOKEN to restart the request.

Continue this process until the system completes processing all the data specified by the request.

To avoid coding separate IXLCACHE invocations with RESTOKEN or EXTRESTOKEN each time you need to restart the request, code a single IXLCACHE invocation with the restart token initialized to all zeros on the original request. Every time you need to restart the request, you can set the restart token equal to the value returned in the CAARESTOKEN or CAAEXTRESTOKEN field of the answer area on the previous request.