Support for a Databricks connection with multiple catalogs

[AlizaminJ]

First of all, I am not quite sure if it is the issue. Hoping for understanding as there is no meaningful log in Metabase admin to supplement the issue here.

We have recently enabled a relatively new feature in Metabase, multiple catalogs within a single Databricks connection. As far as I see it is not supported, as dbt docs are not propagated to Metabase. I tried to adjust with database/schema/model_filter, but no result. It either silently fails, or complains with something like "WARNING:dbtmetabase._models:Table 'BRONZE.MY_MODEL' not in schema 'BRONZE'

Maybe worth mentioning, the same table appears differently depending on the connection type. Upon search, "MY_MODEL" in Metabase console:
MY_MODEL - CONNECTION (CATALOG.SCHEMA) --> connection with multiple catalog, docs not propagated!
MY_MODEL - CATALOG(SCHEMA) --> connection with single catalog, docs propagated!

[gouline]

I don't use it with Databricks, so no current plans to implement this. But if someone is interested in contributing it, happy to review and merge your pull request.

[alxsbn]

@gouline @AlizaminJ Maybe it's not linked to this issue (I didn't have activated the multiple catalog feature). At your opinion, right now I have a Databricks Database named lakehouse setuped with a default catalog like this:

The following dbt-metbase command didn't push the semantic type (I'm testing with one model only):

dbt-metabase --config-path .dbt-metabase/config.yml models --include-models reward_step_completion -v
INFO     2025-10-16 17:18:40 +0200 — Metabase session established                                                                                                                                     metabase.py:61
INFO     2025-10-16 17:18:47 +0200 — Table 'CORE.REWARD_STEPS_COMPLETION' is up to date                                                                                                               _models.py:222
INFO     2025-10-16 17:18:47 +0200 — Field 'CORE.REWARD_STEPS_COMPLETION.ID_CUSTOMER' is up to date                                                                                                   _models.py:433
INFO     2025-10-16 17:18:47 +0200 — Field 'CORE.REWARD_STEPS_COMPLETION.REWARD_LEVEL' is up to date                                                                                                  _models.py:433
INFO     2025-10-16 17:18:47 +0200 — Field 'CORE.REWARD_STEPS_COMPLETION.REWARD_STEP' is up to date                                                                                                   _models.py:433
INFO     2025-10-16 17:18:47 +0200 — Field 'CORE.REWARD_STEPS_COMPLETION.IS_COMPLETED' is up to date                                                                                                  _models.py:433
INFO     2025-10-16 17:18:47 +0200 — Field 'CORE.REWARD_STEPS_COMPLETION.REWARD_STEP_LAST_TRANSACTION_AT' is up to date                                                                               _models.py:433

So everything is flagged as up to date. I tried with new description and the update was well done but not the semantic type

[alxsbn]

@gouline @AlizaminJ Guys I ping you again, could be interesting to have some help on this one

[gouline]

Not related to multi-catalog and I don't have Databricks to test your case against. If you end up figuring out what's breaking, raise a separate issue (or a pull request with the fix).

[AlizaminJ]

@alxsbn , I am not quite sure either if it is related (as I have enabled multi-catalog). But I have managed to push semantic types and other docs during development of this feature 🤞