We have issues with AF Analytics missing essential calculations, because we have peaks in data ingress, leading to peaks in the UpdateManager queue. There are many consumers, but mainly PI Vision and AF Analytics.
We see that we have lost events because the UpdateQueue grows beyond the size of the TotalUpdateQueue tuning parameter. That leads to Analytics missing essential updates (e.g. the start of a batch), but you can't be sure unless you monitor the updatequeue for the individual connections.
Now the problem is that this is not easy to capture in a Performance counter. The UpdateQueue size is in a performance counter created for each connection. And on every reconnection this receives a new ConnectionID.
How now to capture these counters for all the calculation groups of Analytics? Is that possible through performance counters, or is the only option to use the AFSDK access to the DIagnostics? OSIsoft.AF.Diagnostics Namespace