Looking at the source code on develop branch, https://github.com/apache/incubator-predictionio/blob/develop/data/src/main/scala/org/apache/predictionio/data/storage/hbase/HBEventsUtil.scala#L270
when events are exported the eventId is there, so at import time that eventId will be used as rowKey: https://github.com/apache/incubator-predictionio/blob/develop/data/src/main/scala/org/apache/predictionio/data/storage/hbase/HBEventsUtil.scala#L147-L150 So it should replace the existing row because they have the same row key. Same cannot be said if you manually created a replacement event with the same name, entityId, etc. To be sure you can export your events to file, import it to a test appId/channelId twice and see if there's any duplicated events (say you can check using the event server) Tom On Mon, Aug 8, 2016 at 7:09 PM, Jose Rivera-Rubio < [email protected]> wrote: > Hi all, > > *Problem*: > > I'll be generating dumps of my event data using pio export and then > running pio import using these dumps without doing pio app data-delete. > > *Question*: > Is pio import running any duplicity checks or the data will be imported as > is, generating duplicated eventIds? > > Many thanks! >
