PI to CONNECT known issues

The following are known issues in PI to CONNECT:

• If there is no default PI Server associated and listed for your AF server, you will not be able to transfer data to CONNECT data services.

• PI to CONNECT Agent Configuration Utility: Detected Data Archives fail to load if the user does not have permissions to one or more of the referenced Data Archives. The workaround is to add the Data Archive by selecting the green plus icon in the utility and then manually typing the name of the Data Archive.

• Query search results that contain a very large number of PI points (> 1 million) will generate an exception error and not be processed.

• PI points with the type Float16 are supported, but the data is sent as Float32 to the cloud. The initial streaming value is rounded in the historical archive, so historical retrieval of the value may not match the initial value exactly.

• There is a known issue where streaming updates can be lost if the user is deleting events, which is common only when the user is deleting and recalculating events via the PI Analysis service. In this case, we suggest setting the StreamingSendTaskCount parameter in C:\ProgramData\OSIsoft\PIToOCS\appsettings.json to 1 ("StreamingSendTaskCount": 1) so that events are guaranteed to be sent to CONNECT data services in the order they are received.

• The PI to CONNECT Agent requires read access to all AF databases, even databases for which no data is being transferred. Otherwise, the agent will encounter issues with AF indexing, causing the agent to stop collecting data on restart. The error reported in the PI to CONNECT Agent Windows Event Log will be:

  EventId: 191, Failure during AF indexing operation: ProcessChangesAsync
  Exception: System.NullReferenceException: Object reference not set to an instance of an object.

• If an agent is installed on two separate machines and the Agent ID, Tenant ID, and GatewayBaseURI are the same for those two agents, data loss will occur. The cloud creates a single event hub for these two agents because the event hub name is determined by the cluster region, the Tenant ID, and Agent ID. However, two different SDS services read from this event hub and it is indeterministic which service picks up the events, resulting in data loss.

  This issue can be caused by copying the appsettings.json file between machines or by cloning a machine with an installed agent. Resolve by uninstalling and reinstalling the agent on one of the machines to generate a unique Agent ID.