I have a table that has 535M rows (~ 60GB) and so decided to try ODAG on Qlik Sense. The problem: I can't get past the first and most basic step: making a Summary App.
The table structure is as follows:
event_date
geo_city
event_name
device_operating_system
app_info_version
user_pseudo_id
2021-09-14
City A
Event A
Android
3.59.0
123456
2021-09-14
City A
Event X
Android
3.59.0
123456
2021-09-14
City B
Event A
IOS
3.59.1
222444
The original table actually has more fields, but these are the ones that matter and thus the ones I'm gonna use.
By the end of it, I want to know on event_date "A", geo_city "B", using the OS "C" with the app version "D", how many distinct users did the event "E" and some more complexes things, but the point is: the field "user_pseud_id" can be used as count(distinct user_pseudo_id) or just count(user_pseudo_id).
[tb_events]: SELECT event_date, event_name, geo_city, device_operating_system, app_info_version, count(distinct user_pseudo_id) FROM smart_data.tb_events GROUP BY event_date, event_name, geo_city, device_operating_system, app_info_version;
It then says the field user_pseudo_id was not found. But other apps (which we use just a small part of this table) can find this field just fine.
Taking out the distinct and using just count() made no difference. Keeping or not ' ' around the fields and table name also made no difference. Putting the count in the load statement (I was getting desperate) also made no difference.
It also doesn't work trying to load without using count or group by. It then says the field 'event_date' was not found. Again, on other apps with a snippet of this table this field works.
It also doesn't work If I try to leave all fields in the LOAD statement, but use only event_date, event_name and count(user_pseudo_id) in the SELECT statement.
So, what am I missing? How can I do a Summary App? Would it be better in this case to give up on ODAG and try another method?
