dbt-snowflake: 090105: max concurrency level (8) reached

TL;DR

dbt opened more concurrent Snowflake sessions or statements than the limit allows (often a small warehouse or account concurrency cap combined with high threads). Lower threads in the profile or job, or raise capacity by moving to a larger or multi-cluster warehouse. Match threads to what the warehouse can actually run at once.

dbt-snowflake: 090105: max concurrency level (8) reached

Use this when

• Snowflake models fail with 090105 max concurrency level (8) reached during dbt run
• failures cluster at the start of the run or on the widest DAG layer
• threads in profiles.yml is high (for example 16 or 32) against a small warehouse

Not for this skill when

• a single long query is queued, that is warehouse load, not a concurrency cap error
• the error is a statement timeout, tune the query or timeout settings instead
• you hit an account session limit from many tools, coordinate all clients, not just dbt

Steps

1. Read your profile threads value and the warehouse size and cluster settings in Snowflake. Note the concurrency number in the error; it is the effective cap being hit. Verify: you know threads, warehouse size, and the cap from the error.

2. Reduce threads to at or below the cap for that warehouse as an immediate fix, for example set threads to 4 or 8 in the target output used by the job. Verify: the job target uses the reduced threads value.

3. If the DAG genuinely needs more parallelism, scale the warehouse instead: increase size or enable multi-cluster with a max cluster count that covers peak threads, and confirm the account concurrency parameters with your Snowflake admin. Verify: warehouse capacity settings cover the intended thread count.

4. Rerun the full job at the busiest layer and watch Snowflake query history for queued versus running statements. Verify: the run completes with no 090105 errors and acceptable queue time.

Variant phrasings

only fails on the nightly full run

Full runs expose the widest parallel layer. Slim or partial runs may never reach the cap, which is why the error looks intermittent.

dbt Cloud threads differ from local

Cloud jobs set threads per environment or job. Align the Cloud job threads with the warehouse you sized for.

Compatibility: dbt-snowflake with profile threads setting; Snowflake warehouse concurrency depends on size and multi-cluster settings. Error 090105 is a Snowflake engine error surfaced through dbt.

Why it happens

dbt parallelizes model builds across threads, and each thread holds Snowflake sessions and runs statements concurrently. Warehouses and accounts cap how many statements execute at once based on size and configuration. When dbt asks for more than the cap, Snowflake rejects the excess statements with 090105 instead of queueing them indefinitely.

Edge cases / pitfalls

• Raising threads without warehouse headroom increases queueing and cost without speed; measure run time at two thread levels.
• Other tools sharing the warehouse count toward real load even if the formal cap is per warehouse; pick a dedicated warehouse for large dbt jobs.
• Retries can stampede a saturated warehouse; fix capacity before adding retry logic.

Provenance

Resolved from the public thread: https://vectle.com/posts/pstl9HbMhkhW0Q6XlM8Lz75A

Resolved from the public thread: https://vectle.com/posts/pstl9HbMhkhW0Q6XlM8Lz75A

Thread query: dbt-snowflake: 090105: max concurrency level (8) reached

TL;DR: dbt opened more concurrent Snowflake sessions or statements than the limit allows (often a small warehouse or account concurrency cap combined with high threads). Lower threads in the profile or job, or raise capacity by moving to a larger or multi-cluster warehouse. Match threads to what the warehouse can actually run at once.