Not a specific answer to your question, but why the CalculationWaitTimeInSeconds on 600? Default that is 5 seconds, and your setting introduces a delay of 10 minutes to every calculation by default...
Nice catch Roger!
Amit is the lag greater than 10 minutes?
There are many factors that may cause the performance issue with Analysis, like Issue with server performance because of memory and CPU cores in the Analysis machine, Network issues etc.., I advise you to check the logs in PI Server and Analysis Server, Check the Analysis hang dump as per KB00046 - How to get a hang or crash dump and how to obtain and use debugging tools
Let us know about the version of PI Analysis and the AF Client you are using.
Check the Analysis Performance counters as per PI Server , this will give you the clear idea about the Analysis performance. For more details check the below link,
Also, check the below KB articles for your reference,
3 of 3 people found this helpful
Changing a configuration parameter may alleviate the performance issue, however a better approach is to understand why there is a performance issue in the first place. We have customers running hundreds of thousand analyses without any problems, so 2800 is definitely not a lot. The issue here is most likely that an analysis takes longer to evaluate than the frequency at which it is triggered. We typically only change configuration parameters if there is an absolute need.
As a recommendation, is it possible for you to upgrade to AF 2018. In this version, PI System Explorer will give you an indication if the analysis won't be able to keep up:
If the Avg Trigger is faster than the Avg time it takes to evaluate, then the analysis will always fall behind and perpetually lag. The solution for this is to either determine why the analysis is so expensive and fix it , or to decrease the frequency of the calculation. Expensive analyses are often explain by a violation in one or many of our best practices that can be found in this KB: KB01641 - PI Analysis Service Best Practices
Is it possible for you to share the analysis configuration of one of the lagging analyses?
Can you please confirm your AF Client version as we have faced the similar kind of issue and OSIsoft advised us to upgrade the PI AF Client to 2018 version.
We were running into internal WI 189018, and by extension, WI 195013 and OSIsoft confirmed that both of these issues were resolved in PI AF Client 2018, and upgrading the PI AF Client on the machine with the PI Analysis Service running is the only method available to resolve this issue. The one workaround that is available is to restart the PI Analysis Service.
For your information,
WI 189018- This is an issue present in both PI AF Client 2017 and PI AF Client 2017 R2 that causes update consumer calls to dead lock when reconnecting to the PI Data Archive.
WI 195013- This is an issue that would trigger when the update calls dead lock, would cause the problem to cascade to other threads and eventually cause the whole process to hang.
Both are resolved in PI AF Client 2018 version.