OPC interface retries sending InstrumentTag/ItemID to OPC Server
The idea is to have PI Interface for OPC DA/HDA to be able to send again Instrument Tag/Item ID after a configurable time when OPC Server is not able to resolve it (AddItem failed). This would be a workaround for an OPC server issue and would help customers avoid the need to restart interface or modify affected tags in order to have tag incorporated into interface's list.
Hi, we have some OPC DA interfaces connecting to experion OPC servers where we saw following phenomen: Point is collecting data Point is set in out of service by a DCS engineer during service Point is set active again, but does not restart data collection. To resolve this issue we had to update the tag. Also a restart of the interface resolved the issue.
Dataloss is present untill we restart the itnerface or reactivate the point. Automatic reinitialisation of tags (scheduled based) could solve the problem.
David Hillyer commented
This feature is a necessity to ensure the reliable collection of data. I have experienced instances when data was not being collected due to the interface not restarting automatically. The need to manually restart an interface will and has resulted in data loss, which in the pharmaceutical industry is a huge compliance issue.
Because this can prevent data loss, it should be a high priority. Data loss will be caused without this feature whenever an OPC server loses it's health, and sends all data to the PI interface marked with BAD status. This feature request will prevent extended periods of data loss, in cases where an unhealthy OPC server marks all data from a DCS system as bad. The OPC/UA interface has a default setting that requires 80% of data from each interface to be of good quality, or else it marks the interface as being in "CONNECT/NODATA" state. That setting can be lowered, but if it is lowered to 0, it will cause excess resource usage at the PI interface hardware and network level. The percent good quality limit, if not reached, will force the interface to shutdown, despite interface redundancy. Because the bad tags are not ever rechecked, the interface does not restart again even after all data returns to good status when the OPC server returns to health. This delay in PI interface restart can cause data loss over large numbers of PI tags, and the period of data loss can be extended because problem notifications and interventions are required before the interface can be restarted.