KB38972: In MicroStrategy Web 9.x, documents based off view reports run against the intelligent cube, even though there is an existing valid document cache.

SYMPTOM:
 
In Strategy Web 9.x, when a user executes a document based off intelligent cube reports, when the "Create cache for all requested incremental fetch blocks, group-by and control element selections" option is checked, caches get created successfully and any selector changes or document execution within the same session hits existing cache successfully. However if the user logs off their web session or navigates to a different document, and then re-visits the above document again, for the 1st selector change, the document job will run against the intelligent cube and will not hit the existing cache.
 
STEPS TO REPRODUCE: 

• Create an intelligent cube with Region and Year attributes and profit and revenue metrics
• Create a report from the above intelligent cube and insert Region and Year attributes and profit and revenue metrics
• Create a dashboard and insert above report as dataset and insert selectors for "Year" and "Region" attributes (doesn't matter whether you use filter or slice)
   
  

• Save document and edit in Strategy Desktop and go to Format, then Document Properties and navigate to "Caching" and check "Enable document caching" > "XML/Flash"  and check "Create cache for all requested incremental fetch blocks, group-by and control element selections" option
   
  

• Run the document in Strategy Web and make selector changes for Year and Region and see that a cache will be created for each selector manipulation or combination.
   
  

• Exit Strategy Web and log back in and run the document again. The document hits the existing document cache and the hit count against the Intelligent Cube does not go up.
• Now, change the selector to any value (document is cached for all selector values). It is seen that the document does not hit the existing document cache and runs against the Intelligent Cube.
• Change the selector to a different value. In this case, an existing document cache is used and does not run against the Intelligent Cube.

 
CAUSE:
This is working as designed.
 
When the document is executed for the very first time i.e. when no document caches exist, it results in a full document execution and a document instance gets created in memory. Following this execution, when the first selector change is made, since this is a server manipulation, a document instance needs to be created. Since the document instance from the initial execution still exists in memory, a new document instance does not need to be created.
 
When the document is executed for the 2nd time,  (when previously created document caches exist) the document runs against the cache. Since this is a document cache hit, no document instance is created. Now, when the first selector change is made, since this is a server manipulation, a document instance needs to be created. Hence, a full execution needs to be performed against the Intelligent Cube. Now, for subsequent selector value changes, since a document instance already exists in memory, a full execution does not need to be performed and existing document caches are used.
 
 
ACTION:
 There is no action to be taken as this is working as designed.