I imported an unknown number of events into my PIO eventserver and now I want to know that number (in order to measure and compare recommendation engines). I could not find an API for that, so I had a look at the MySQL database my server uses. I found two tables:
mysql> select count(*) from pio_event_1;
+----------+
| count(*) |
+----------+
| 6371759 |
+----------+
1 row in set (8.39 sec)
mysql> select count(*) from pio_event_2;
+----------+
| count(*) |
+----------+
| 2018200 |
+----------+
1 row in set (9.79 sec)
Both tables look very similar, so I am still unsure.
Which table is relevant? What is the difference between pio_event_1 and pio_event_2?
Is there a command or REST API where I can look up the number of stored events?
You could go through the spark shell, described in the troubleshooting docs
Launch the shell with
pio-shell --with-spark
Then find all events for your app and count them
import io.prediction.data.store.PEventStore
PEventStore.find(appName="MyApp1")(sc).count
You could also filter to find different subsets of events by passing more parameters to find. See the api docs for more details. The LEventStore is also an option