Metabase dropping Views unprompted in DB - no idea why

Hi all. Been dealing with a situation this week where a handful of seemingly random DBT views are being dropped. Searched up and down my stack before eventually finding this line in some of the Metabase logs:

[22b9f3f1-715a-40e6-94ea-019ec03a9a07] 2024-06-19T14:52:00-07:00 INFO metabase.sync.sync-metadata.tables Marking tables as inactive: (Table 'marts.dim_tenants' Table 'export.export_hubspot_tenants' Table 'marts.agg_tenant_sessions' Table 'int.int_integration_operation_history' Table 'marts.dim_compa_user_list' Table 'marts.fct_mixpanel_user_sessions' Table 'marts.alert_3_duplicates' Table 'export.export_hubspot_join_keys' )

I'm at kind of a loss here. These tables were built by an overnight DBT job. Some of them are pulled into Metabase as models, some aren't used at all. We're a very small company with low Metabase usage overall, so I highly doubt it was anyone playing around with something in Metabase. Has anyone seen this before? I want to keep sync on as I keep adding more tables/models/content, but I don't see any other posts or options on not to drop database objects

that's done automatically by Metabase if it can't find the table. Do you know if those tables exist while Metabase is running the daily sync process?

I can see in the DBT logs they existed as of 5am. I turned off additional Postgres logging today but not aware of any DBT jobs that have run between now and then, or Users querying the DB today.

Can you confirm that the tables exist now? Can you query them from the native SQL editor in Metabase? If so do they still fail to come in with a manual sync?

They exist now after a nightly DBT run. Happened again overnight but didn't reboot my DB instance in time for full DDL logging to take effect. Still a mystery... my current theory is perhaps Metabase is running a sync for drop down values, and queries a VIEW during a time when a parent DBT model is being recalculated (and thus the model being sync'd would be 0)