Hi Team. Is there a way to disable initial indexing/metadata queries in a cluster level or during TML import? The issue we are facing right now is we are creating many orgs with lots of wide tables(105 and most table has >200 columns and two tables have >400 columns) in a day for testing and initial indexing is consuming lot of snowflake credits. We already have disabled indexing in cluster level but initial table import is not respecting that config and hence lots of queries are fired to either get metadata or sample value or for internal thoughtspot use.

message has been deleted

Short answer: there is no documented switch in TML import to suppress all initial metadata/introspection work. From the official docs, TML import supports options like

skip_diff_check

, import policy (

PARTIAL

,

ALL_OR_NONE

,

VALIDATE_ONLY

,

PARTIAL_OBJECT

), org context, and **skip CDW validation checks for Table TMLs**—but nothing documented as “disable initial indexing/metadata sampling entirely during import.” What is documented and relevant to your case: 1. Use Table TML import options that skip CDW validation checks (where applicable) to reduce some warehouse calls during import. 2. Set

skip_diff_check: true

to avoid redundant diff processing when repeatedly importing unchanged objects. 3. Batch dependent objects in one import call (Model + dependent objects) to reduce repeated processing cycles. 4. Use async TML import for scale/throughput (

/api/rest/2.0/metadata/tml/async/import

) so imports can be orchestrated more efficiently and monitored. Given your workload (many orgs + very wide tables), the remaining metadata/sample/internal queries you see during first-time object creation are typically part of required object validation/materialization and may still occur even if cluster-level indexing is disabled. If you want, I can help you build a low-credit import pipeline pattern (import ordering + flags + retry/status flow) tailored to multi-org Snowflake testing.

Could you please ask this in the in-app chat support, the correct team should be able to help you. If you feel this is a bug please create a support case.