Action required: Databricks query history access change

Summary — Databricks is rolling out a change in late June 2026 that masks SQL query text in the query_history system table behind a new Unity Catalog access policy. Monte Carlo's Databricks collector reads this column. To avoid
losing Monte Carlo functionality, you must add Monte Carlo's service principal to a new Databricks group (databricks_pii_access) before the change activates in your environment.

What's changing

Databricks is updating the query_history system table so that the statement_text column (the SQL text of each executed query) is masked by default. Going forward, only members of a new Databricks group, databricks_pii_access, will be able to read query text.

This is a Databricks platform change, not a Monte Carlo change. The reason, per Databricks: SQL query text often contains sensitive data, including PII.

When this takes effect

Databricks will ship this change in a Databricks Runtime (DBR) release in late June 2026. It activates in your environment when that runtime is rolled out to your workspace.

Databricks is rolling this out in two steps:

1. Step 1 — Group introduced (no impact yet). The databricks_pii_access group becomes available in your Databricks account. Masking is not yet active. Use this window to add the Monte Carlo service principal (and any of your own
   users who need query text access).
2. Step 2 — Masking activates. Only members of databricks_pii_access can read statement_text. If Monte Carlo's service principal is not in the group at this point, Monte Carlo will stop receiving query text from your environment.

⚠

Do not wait for Step 2 to act. Once masking activates, any Monte Carlo features that depend on query text will silently stop working until access is granted.

What you need to do

1. Add Monte Carlo's service principal to databricks_pii_access

Before Step 2 activates in your environment:

1. Identify the service principal that Monte Carlo uses to authenticate to your Databricks workspace. This is the same service principal you configured when connecting Databricks to Monte Carlo. You can find it in Monte Carlo → Settings →
   Integrations → Databricks (under the connection's service principal field).
2. In your Databricks account, add that service principal to the databricks_pii_access group. See the Databricks documentation on managing groups for the exact steps.

2. Verify Monte Carlo collection after Step 2

Once Step 2 activates in your environment, confirm that Monte Carlo is still receiving query history from Databricks:

1. In Monte Carlo, open any Databricks table.
2. Go to the Queries tab.
3. Confirm that recent queries are appearing.

If queries stop appearing after Step 2, the service principal is likely not in the databricks_pii_access group. Re-check the group membership in Databricks.

What's affected if Monte Carlo loses access to query text

The following features rely on query text and will be impacted if Monte Carlo cannot read statement_text:

• Query Performance monitors and Query Performance Anomaly alerts
• Query Insights and the Performance page
• Field usage / column popularity signals
• Table activity (read/write counts, queried-tables detection)
• Impact Analysis users on incidents (paginated user lists with per-user query counts)
• Query Log Details drawer (the per-query drill-down across Monte Carlo)
• Discover Queries widget
• AI monitor recommendations that depend on query patterns

Frequently asked questions

Do I need to do this if my Databricks runtime is not upgrading right away?
The change activates per environment as the runtime upgrade reaches it. Add the service principal now so you are covered whenever your workspace's runtime is updated.

Will my own users still be able to see query text in Monte Carlo?
Yes. Monte Carlo's access to query text in Databricks does not change which Monte Carlo users can view query text inside Monte Carlo. Monte Carlo's permission model is unchanged.

Where can I confirm the group name and Databricks' exact rollout details?
Refer to Databricks' communication on the Unity Catalog query history masking change, or contact Databricks support.

Does this mean PII is exposed in Monte Carlo?
No, Monte Carlo does PII filtering on the collection side.

Need help?

If you have questions, contact your Monte Carlo account team or open a support request.