Forum Discussion

koenverbeeck's avatar
koenverbeeck
Advocate II
10 years ago
Solved

Issue with multiple SSAS cubes (the visual has unrecognized fields)

We have a SSAS MD database with multiple cubes.

An enterprise gateway is installed and configured. The account is a SSAS server admin.

 

We created a test report in Power BI Desktop on one of the cubes and published it to the Power BI service. Everything works as expected. Then I created another report in Power BI Desktop, on one of the other cubes. Published it succesfully to the service.

 

However, this time not everything works. Measures or dimensions unique to the second cube are not displayed. If I add them to a visual, I get a big cross with the message "The visual has unrecognized fields". But, if I use a conformed dimension which is also in the first cube, it does work. The field list displays all of the metadata, but Power BI can't display the  data unique to that cube.

 

When I monitor with profiler, nothing arrives at the cube. If I use a shared dimension, the query does arrive and is processed correctly. The query is addressed to the correct cube.

 

It seems like Power BI is messing up to the models (since shared stuff works), but when I use something that works, the query is sent to the correct cube.

 

Any ideas?

  • orionlee's avatar
    orionlee
    10 years ago

    The fix for this issue has been deployed to production (should have arrived ~mid-June).

     

    If you are still having problems with this issue please let us know, so that we can find out if the issue has regressed again since then or if the fix did not cover your particular scenario.

14 Replies

  • devastate's avatar
    devastate
    Regular Visitor

    Since a few days this problem has re-appeared for one of our customers. It behaves exactly as described by Koen Verbeeck, including the "switching" working reports between days.

     

    Can this issue be reopened?

    • jabo187's avatar
      jabo187
      New Member

      Since wednesday we are experiencing the same problem.

  • Update: when monitoring the gateway server with fiddler, the JSON response to a non-working dimension had the following error:

     

    Could not resolve model references in the Semantic Query. Affected items: [dimension name].

    • koenverbeeck's avatar
      koenverbeeck
      Advocate II

      OK, so I tried again this morning, after the weekend, and suddenly everything of the second cube works. Without any intervention. However, now some visualizations from the first cube - who did work - are now not working. Weird.

      • orionlee's avatar
        orionlee
        Microsoft Employee

        Thanks for raising this issue koenverbeeck, we have looked into it and identified a caching issue in the PowerBI service for cases where there are multiple cubes for the same database.

         

        A bug has been filed for the feature team. In the meantime, unfortunately there are not any workarounds we can suggest for the problem other than waiting 10-15min for the cache to clear in the service before opening the report.

         

  • I' suffering the same issue.

    my first cube in my OLAP database is working fine, the second one has the same error.

    and I suffer this problem with both an OLAP and Tabular instance on my server.

     

    • orionlee's avatar
      orionlee
      Microsoft Employee

      The fix for this issue has been deployed to production (should have arrived ~mid-June).

       

      If you are still having problems with this issue please let us know, so that we can find out if the issue has regressed again since then or if the fix did not cover your particular scenario.

      • koenverbeeck's avatar
        koenverbeeck
        Advocate II

        Hi,

         

        somehow I completely missed your reply.

        I just checked and everything seems to be working.

         

        Thanks for the fantastic support!

    • koenverbeeck's avatar
      koenverbeeck
      Advocate II

      The original question was solved over a year ago.

      The problem was clearly described as a caching bug, so the services were indeed running.