Option to limit tables/fields that are synced to specific schemas

[parkotron]

We're working on a relatively small system in redshift that uses a shared redshift cluster. There are lots of schemas most of which are not ours (and we don't have permission to read) but our local postgres is becoming unnecessarily heavy due to the automatic caching of fields in all schemas into metabase_field (it would likely be a few hundred rows, but is getting towards a hundred thousand).

A possible enhancement would be to have the option to limit field caching to schemas that have been viewed rather than crawling all schemas up front.

⬇️ Please click the 👍 reaction instead of leaving a +1 or update? comment

[rafaelveloso]

Being able to configure which schemas we want to be scanned by metabase would be very helpful. In PostgreSQL we can define permissions per schema but that doesn't prevent any role from seeing all schemas tables structure. If we have multiple schemas in one database and create a specific role (to be used by metabase) with read permissions in one schema, it won't be able to pull data from any schema other than the one it has permissions to, but metabase will still sync metadata for all of them.

[MrMauricioLeite]

Same problem here. Our redshift schemas show up but I can only run querys on my schema.

Other thing that sucks is that all those schemas have lots of tables and we have to hide one by one, since there is no "Hide All" option anywere.

[robinfrick]

+1 this would be taking #2086 a step further. For me hiding would already be a great step forward. Completely disabling the syncing of the schema is even better 👍

[justin-yan]

Would love to see this as well. We create a new schema when we load data into our warehouse so that we can atomically publish the new day's data all at once by doing an ALTER SCHEMA RENAME TO. We then keep the old schemas around for a few days with a date suffixed so that we have an audit trail.

Hiding would actually not solve our problem, since each audit schema has a unique suffix, but being able to limit imports to specific schema names would be a big usability win for us!

[toybaelescione]

An option to pick which schema to be scanned by metabase will be very helpful for us as well. We have Oracle for our Data Warehouse and we have hundreds of schema and hundred thousands of tables, and not all tables are used for analysis.

[Update on Jun 6, 2018]
We have Metabase runs on PROD, and we think it is wonderful, but we need to kill the service then turn it on, every single day. Due to very high buffer gets on our Oracle DWH that came from queries from Metabase sync job.

The database killer query was running on top of view that contains complex queries. Unfortunately, deleting the view is not one of our option now,

[camsaul]

Fixed by #10139