Skip to main content
The Cognite PI extractor can’t synchronize PI content to CDF if the time series have invalid, incorrect, or missing data points. Use the Cognite PI replace utility to re-ingest time series to CDF by optionally deleting a data point time range and ingesting the data points in PI for that time range. By default, the PI replace utility re-ingests the entire time series, but you can limit this to a time range with the start-time and end-time parameters in the configuration settings. You can also configure the utility to react to data loss and reconnection incidents detected by the PI extractor.
The PI replace utility doesn’t modify the time series in CDF, only the data points. It assumes that all configured time series exist in PI and CDF and will skip any time series in PI that it doesn’t find in CDF.
Follow the PI replace progress through logs and real-time metrics and see the execution operation summary report.

​
Typical sources of inconsistencies in data points

  • Downtime on the PI extractor or loss of data between the PI AF SDK client and the PI server. The PI extractor can monitor these incidents.
  • The PI extractor reconnects to the PI server, for instance, when the extractor recovers from failures. During reconnection, the extractor may miss the notification from PI about historical updates or insertions of data points, which can cause missing or incorrect data points in CDF.
  • The client-bound queue in the PI server lost data. The typical cause is queue overflow due to bursts of data inserted and the client failing to fetch it. The dropped data never reaches the PI extractor, which can cause missing or incorrect data points in CDF.
  • Other sources of inconsistencies can be bugs in the PI extractor, causing incorrect data points in CDF.