Detector makes use of Xmanager to route requests from one system in a Sysplex to another.
There are 2 Xmanager parameters (TIMEOUT and CPUTIME) involved in a cross system
view of a datastore as described above.
TIMEOUT specifies the number of seconds that one Xmanager will wait for execution,
of a request routed to another Xmanager in the Sysplex complex, to complete before
signaling a timeout error.
The default value for TIMEOUT is 20.
CPUTIME specifies the number of CPU seconds the Xmanager lets a process
consume when satisfying a request routed from another Xmanager in the Sysplex
The default value for CPUTIME is 10.
When a cross system request is routed across from a local Xmanager to a remote
Xmanager, the remote Xmanager allows the routed request the specified CPUTIME
to complete. If the request does not complete within this time then a U0966
abend is registered in that remote Xmanager.
The net effect is that the request will eventually exceed the TIMEOUT value
in the local Xmanager, at which point the DT663E is seen in Detector.
In this case, one or both of CPUTIME and TIMEOUT values may need to be increased
in order to satisfy the request. CPUTIME should generally be increased first if the remote
Xmanager has registered a U0966 abend related to the request.
As a guide to how much the value may need to be increased by, attempt to access the
datastore locally from the lpar on which it resides. The time taken should give an
indication of the value that CPUTIME will need to be set to in the Xmanager.
If the increased value for CPUTIME now exceeds the TIMEOUT value then you should
also look to increase the TIMEOUT value to at least the same value as being used
The CPUTIME and TIMEOUT values do not factor in any additional time that it may take
to recall the underlying datasets for a datastore, if they have been previously archived.
So this should be a consideration when setting the values for the 2 parms, particularly if the
expectation is that datastore datasets will more likely be archived than not.