dbt recreates the whole table?

The problem I’m having

Not a problem but more of we want to understand what’s happening. We noticed that dbt recreates the whole table and processed 41TB of data instead of just running it for a specific processing date.

What we noticed:

  1. dbt runs our models and create it under model_name__dbt_tmp
  2. dbt recreates the whole table by running; the table was already existing and had a lot of data. Because of this it processed 41TB of data
create or replace table model_name as (
 select
    col1,
    col2,
from model_name
);
  1. dbt runs partition merges into the same table

Some settings we have;

  1. incremental matieralization
  2. require_partition_filter = true
  3. +on_schema_change: "sync_all_columns"

Question: Why is it doing step 2 on what I listed above?

Thank you!