^Seem to be close to having this working, however ...
# integrate-tableau-datahub
n
^Seem to be close to having this working, however I am now noticing that a bunch of datasets were generated by the Tableau ingestion. These seem to be parsed from the Custom SQL, however the databases are off, it seems that because the Custom SQL doesnt use a fully qualified table name ex. select * from `analytics.table1`; datahub generates a dataset called
analytics.analytics.table1
and it cant figure out how to connect that Custom SQL to the actual snowflake Table which is
database1.analytics.table1
. How can I help it along here? Id rather it not map upstream at all than just guess and generate a dataset that sits in datahub. Ideally id get it to map back to its actual snowflake table (which is already in datahub).
@modern-artist-55754 Any insight on this? I see this in datahub docs but it makes me a little nervous given the description...
Given I dont really understand how it is "deriving" the database for the generated URN.
m
This is a major issue with tableau metadata api, it returns very inconsistent metadata (partly because how the connection is setup in tableau). Sometime datahub has to guess, in this case it took the schema name as database name, there is no guarantee on this. I have not used the lineage override feature, but i suppose you can do database_override_map with `analytics1`:`database` and see how it goes. Alternatively, you have to go to tableau and change the database connection to connect to specific database rather than relying on default database.
a
Hi! Thats exactly what you need. More context here
n
Update on this, the map wasnt possible cuz we have schemas that have the same name in multiple DB's and creating a hardcoded map was not sufficient. Our Analytics team is actually deprecating Custom SQL, so we dont have to deal with it, which is great, but not a fix for the fact that the Tableau API is faulty.