Issue like this require thorough investigation. It might even be necessary to take and analyzing some hang dumps to understand what's exactly happens. The forums are not the right place to troubleshoot such issues and hence I recommend working with regular Technical Support. For sure it is useful to update this thread later with the information of what has caused your issue.
The first things to look at is the configuration of your Analysis output and I would also check if there's a Custom Data Reference involved which potentially can cause issues like this. You are welcome to share this information here but I doubt, it will be sufficient to make a good guess of what's going wrong.
When you map an Analysis to an Attribute which is not using the PI Point Data Reference, results will not be persisted to a PI Point but evaluated client side on request. This means that the required raw data must be pulled by the client each time analysis results are used and depending on the analysis, this can be an expensive operation. This can result into NullReferenceException in case resources become exceeded but there are other possible reasons too. I've looked at the NullReferenceException details but didn't recognize something ringing a bell for me.
1 of 1 people found this helpful
The first thing I'd recommend is making sure you have at least two CPU's / virtual processors. As long as you have memory-consuming processes like SQL Server and PI Analysis Service on the same box you should definitely make sure you have multiple processors to distribute the load.
As Gregor mentioned, analyses mapped without history can make the SQL Server process CPU usage skyrocket.
I would echo Asle's advice about re-configuring your VM to be at least 2 cores. Try that, and if the problems persist, then call Tech Support.
Thank you all for the good suggestions and the quick response!
After further investigation i found some deleted (referenced) elements and anlyses that for some unknown reason was stuck in the PIFD database.So I exported everything I needed from AF using System Explorer and recreated the PIFD database (using the C:\Program Files\PIPC\AF\SQL\GO.bat script). Then I restored back the AF content, and the problems seem to be gone .
I think the root cause were some circular reference between deleted elements that was stuck in PIFD without appearing in System Explorer.
Thanks again and best regards,
If you still have a backup of the system when it was in that state and want to confirm it was a circular reference issue, please submit it to tech support and I take a look at it. A circular reference might have caused a short spike but it should not have caused it to stay stuck at that level.
AF SQL Dev