Snowflake
What is Snowflake?
Snowflake is a cloud-based data warehouse tool that offers organizations a flexible and scalable storage system. It is great for hosting data that can then be queried and accessed by a business intelligence solution.
Snowflake is provided as a self-managed service that runs completely on cloud infrastructure. This means that all three layers of Snowflake’s architecture (storage, compute, and cloud services) are deployed and managed entirely on a selected cloud platform.
Why Connect Snowflake to Monte Carlo?
Connecting your Snowflake account to Monte Carlo, and giving us access to the data within is essential for Monte Carlo to function. Monte Carlo needs to be able to collect the metadata of your tables in snowflake to enable our Freshness and Volume Table monitors. With access to the query logs of those tables, we will be able to build table and field lineage. With the ability to run SQL queries on those tables, Monte Carlo can run advanced monitors to analyze the data in those tables.
Your agents and analytics depend on this data, so the biggest step in getting value from Monte Carlo is letting it observe your data!
After connecting Snowflake here is what you can expect to see:
Freshness and Volume:
Freshness and volume are monitored by a Table monitor. Connecting the integration doesn't monitor any tables on its own: create a Table monitor and choose the databases and schemas it covers. By collecting metadata on your tables, Monte Carlo’s machine learning algorithm can set automatic thresholds to alert when a table has gone too long without an update, and when an abnormal amount of rows are added or deleted.
Query Logs:
When giving access to query logs Monte Carlo is able to show the amount of read and write queries on a table, and give you access to the details of those queries. Monte Carlo can even highlight when a query appears to have failed.
Table Lineage:
Monte Carlo is able to map the connections between your tables, and can show upstream and downstream connections. We can even connect anomalies that are detected in upstream tables to anomalies in downstream tables.
Tag Ingestion
Monte Carlo has the ability to ingest tags that are placed on Snowflake tables. Tags are displayed on the tables General Information page. By default tag ingestion is not enabled, but can be turned on directly in the UI by enabling "Asset tags collection" on the warehouse integration settings page.
Monitor & Lineage Support
Below are the monitors and lineage support for the Snowflake integration today. Please reach out to your Monte Carlo representative if you have requirements for additional monitors.
| Category | Monitor / Lineage Capabilities | Support |
|---|---|---|
| Table Monitor | Freshness | ✅ |
| Table Monitor | Volume | ✅ |
| Table Monitor | Schema Changes | ✅ |
| Metric Monitor | Metric | ✅ |
| Metric Monitor | Comparison | ✅ |
| Validation Monitor | Custom SQL | ✅ |
| Validation Monitor | Validation | ✅ |
| Job Monitor | Query performance | ✅ |
| Lineage | Lineage | ✅ |
More information on monitors in Monte Carlo.
*We do not support custom monitors on fields and tables named with Snowflake reserved words.
Support for observing Snowflake components
Views: Monte Carlo supports Views — they show up as assets, have opt-in automatic and custom monitors, and are in lineage
External Tables: Monte Carlo supports External Tables — they show up as assets, have both automatic and custom monitors, and are in lineage
Dynamic Tables: Monte Carlo supports Snowflake Dynamic Tables — they show up as assets, have both automatic and custom monitors, and are in lineage
Streams: Monte Carlo supports Snowflake Streams — they show up as assets, have schema-change and custom monitors, and are in lineage.
Tasks: Monte Carlo supports Snowflake Tasks, including task run observability and real-time task failure notifications. See Snowflake Tasks.
Connecting Snowflake
Now that we know the value connecting Snowflake to Monte Carlo can bring, here are the necessary steps to set up the integration.
Admin credentials requiredTo make the read only service user for Monte Carlo, you will need permissions to use the
ACCOUNTADMINrole on Snowflake.
This guide explains how to create a read-only service user for Monte Carlo on Snowflake.
To review all steps necessary to integrate a data warehouse with Monte Carlo, please see here.
Connect end to end with Claude or TerraformWant a guided setup? With the Monte Carlo Agent Toolkit, run the following command and Claude walks you through every step:
/monte-carlo-onboarding connect SnowflakeSee Providing credentials with Claude for guides that fit your setup.
Want it all as code? With the Monte Carlo Terraform provider, one
terraform applycreates the credentials, warehouse and connection, and can deploy a collection agent for them too. See Providing credentials with the Terraform provider or the Connect a Snowflake warehouse registry guide.Prefer the UI? Follow the steps below, then provide the credentials in the onboarding wizard.
Whichever you choose, Monte Carlo needs a Snowflake service user with a key pair. Create it with the script below, or have Claude include it in the Terraform it writes (example).
Creating a service user
Please use the SQL snippet below to create a service user for Monte Carlo. Before using the snippet, you will need to populate the following:
<your_database>: Snowflake database to be monitored by Monte Carlo.
Please note:
- This script will create a Service User that requires a private key for authentication. After running this script to create the user and permissions, you will need to create a private and public key then modify the user to assign the public key.
- If you need to monitor multiple databases in your Snowflake account, please run the snippet multiple times, changing the value of
<your_database>on each run. - Monte Carlo will use an XS warehouse by default. If you expect heavy monitoring workloads, you may consider a larger warehouse, or assigning Monte Carlo's service user to an existing larger warehouse in your account. This setting can be modified later if necessary.
- We recommend using Snowflake's worksheet interface to run the snippet. Please copy the code, make any necessary modifications and use the All Queries option to run it in full.

Running the full snippet using Snowflake worksheets
Row Level SecuritySnowflake allows Row Level Security that can restrict the rows a SELECT query can return. In order to ensure that Monte Carlo monitors are able to work correctly, the Monte Carlo Service user's role needs to be exempt from Row Access Policies. If you notice a monitor's query is failing in the Monte Carlo UI, but succeeding when run by a different user in snowflake, you can use these steps to check for Row Level Security.
Schema vs Database Future GrantsThe script below grants on future objects at the database level only. A database level future grant never overrides an existing grant, so running the script cannot remove access from your existing roles.
Snowflake resolves future grants per object type, and a schema level grant takes precedence over a database level one. From Snowflake's docs:
"When future grants are defined on the same object type for a database and a schema in the same database, the schema-level grants take precedence over the database level grants, and the database level grants are ignored. This behavior applies to privileges on future objects granted to one role or different roles."
A database level grant is therefore ignored in any schema that already has its own future grants for the same object type. Check whether you have any before running the script — the next section shows how.
For a worked example of this behavior, see Snowflake's community article Behavior of future grants when defined at both database and schema level.
Before you run: check your existing future grants
Snowflake applies a database level future grant throughout the database, except in schemas that already have their own future grants for the same object type. Checking first tells you whether the setup script is sufficient on its own.
Generate one statement per schema in the database you plan to monitor:
USE ROLE ACCOUNTADMIN;
USE DATABASE <your_database>;
SELECT 'SHOW FUTURE GRANTS IN SCHEMA "' || REPLACE(catalog_name, '"', '""')
|| '"."' || REPLACE(schema_name, '"', '""') || '";' AS run_these
FROM information_schema.schemata
WHERE schema_name <> 'INFORMATION_SCHEMA'
ORDER BY schema_name;Run the generated statements and note any schema that returns rows.
- No schema returns rows. This is the common case. Run the setup script below and you are done.
- Some schemas return rows. Those schemas have their own schema level future grants. Run the setup script below, then follow "If you found schema level future grants" to add matching grants for those schemas only.
The setup script is safe to run either way — a database level future grant cannot override anything. The difference is coverage: without the extra grants, Monte Carlo will not see objects created later in the schemas that returned rows.
-- Configuration
set mc_username='MONTE_CARLO';
set mc_warehouse_size='XSMALL';
set mc_warehouse_name='MONTE_CARLO_WH';
set mc_role_name='MONTE_CARLO_ROLE';
set database_to_monitor='<your_database>';
-- Set role for grants
USE ROLE ACCOUNTADMIN;
-- Create warehouse for Monte Carlo's monitoring workload
CREATE WAREHOUSE IF NOT EXISTS identifier($mc_warehouse_name) WAREHOUSE_SIZE=$mc_warehouse_size INITIALLY_SUSPENDED=TRUE
AUTO_SUSPEND = 5 AUTO_RESUME = TRUE;
-- Create the role Monte Carlo will use
CREATE ROLE IF NOT EXISTS identifier($mc_role_name);
-- Create Monte Carlo's service user and grant access to role
CREATE USER IF NOT EXISTS identifier($mc_username) DEFAULT_ROLE=$mc_role_name TYPE=SERVICE;
GRANT ROLE identifier($mc_role_name) TO USER identifier($mc_username);
-- Grant permissions to use the new warehouse
GRANT OPERATE, USAGE, MONITOR ON WAREHOUSE identifier($mc_warehouse_name) TO ROLE identifier($mc_role_name);
-- Grant privileges to allow access to query history
GRANT IMPORTED PRIVILEGES ON DATABASE "SNOWFLAKE" TO ROLE identifier($mc_role_name);
-- Grant metadata privileges to database to be monitored
GRANT USAGE,MONITOR ON DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT USAGE,MONITOR ON ALL SCHEMAS IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
-- Grant reference privileges to database to be monitored
GRANT REFERENCES ON ALL TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT REFERENCES ON ALL VIEWS IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT REFERENCES ON ALL EXTERNAL TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT MONITOR ON ALL DYNAMIC TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT REFERENCES ON FUTURE TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT REFERENCES ON FUTURE VIEWS IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT REFERENCES ON FUTURE EXTERNAL TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT MONITOR ON FUTURE DYNAMIC TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
-- Grant read-only privileges to database to be monitored
GRANT SELECT ON ALL TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT SELECT ON ALL VIEWS IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT SELECT ON ALL EXTERNAL TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT SELECT ON ALL STREAMS IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT SELECT ON ALL DYNAMIC TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
-- Grant privileges on objects created in the future
GRANT USAGE,MONITOR ON FUTURE SCHEMAS IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT SELECT ON FUTURE TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT SELECT ON FUTURE VIEWS IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT SELECT ON FUTURE EXTERNAL TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT SELECT ON FUTURE STREAMS IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
GRANT SELECT ON FUTURE DYNAMIC TABLES IN DATABASE identifier($database_to_monitor) TO ROLE identifier($mc_role_name);
If you found schema level future grants
Complete this only for the schemas that returned rows in the check above. Skip it entirely if none did.
For each of those schemas, look at the grant_on column to see which object types have schema level future grants. Add a matching grant for Monte Carlo for each object type present. For example, when a row reports a table object type:
GRANT SELECT ON FUTURE TABLES IN SCHEMA <your_database>.<schema> TO ROLE MONTE_CARLO_ROLE;
GRANT REFERENCES ON FUTURE TABLES IN SCHEMA <your_database>.<schema> TO ROLE MONTE_CARLO_ROLE;Repeat for each object type present in the output: views, external tables, streams, and dynamic tables. Use MONITOR in place of REFERENCES for dynamic tables, and note that streams only need SELECT.
Only grant on object types that already appear in the outputAdding a schema level future grant for an object type that has none makes Snowflake ignore every database level future grant for that object type in that schema, including grants held by your own roles. Check the output for each schema before granting.
Creating a key-pair for the service user
Follow Snowflake's key-pair authentication guide to create a private and public key for the service user you just created.
- The private key will be needed during onboarding in the Monte Carlo UI.
- Monte Carlo supports using an encrypted or unencrypted private key.
- The public key will be assigned to the service user in Snowflake
Use this SQL command to assign the public key to the service user in Snowflake:
ALTER USER MONTE_CARLO SET RSA_PUBLIC_KEY='<public key>';
Exclude the public key delimiters in the SQL statementWhen you generate the public key the output will be in PEM format and look like:
-----BEGIN PUBLIC KEY----- MIIBIj... -----END PUBLIC KEY-----In the SQL command do not include the delimiters, it will look like:
ALTER USER MONTE_CARLO SET RSA_PUBLIC_KEY='MIIBIj...';
External OAuth
Monte Carlo is also able to use an external OAuth Integration that is set up in your Snowflake environment (https://docs.snowflake.com/en/user-guide/oauth-ext-overview). Monte Carlo is able to use either the Client Credentials or Password OAuth Grant flows. Monte Carlo does not work with Snowflake's built-in OAuth service.
This Method requires you to have an existing external OAuth Integration set up in your Snowflake environment. See https://docs.snowflake.com/en/sql-reference/sql/create-security-integration-oauth-external
To configure Monte Carlo to use OAuth to connect to your Snowflake environment, under Settings-> Integrations, in the Snowflake integration creation or edit view, you must provide the following
- Access Token Endpoint - This is the endpoint that MC will use to acquire access tokens for Snowflake
- Grant Type - Must be either Client Credentials or Password
- Client ID - The client identifier issued to the client during the OAuth registration process
- Client Secret - The client secret issued to the client during the OAuth registration process
- Scope - For Snowflake OAuth, scope is used to limit the roles that the Snowflake OAuth user can take. Snowflake requires scope to be set. (https://docs.snowflake.com/en/user-guide/oauth-ext-overview#scopes)
If using a Password Grant Type, you must also provide
- Username - Username of the OAuth User
- Password - Password for the OAuth User
The Snowflake user that MC will use depends on matching the attribute set in EXTERNAL_OAUTH_TOKEN_USER_MAPPING_CLAIM to the attribute set in EXTERNAL_OAUTH_SNOWFLAKE_USER_MAPPING_ATTRIBUTEin your Snowflake External OAuth Integration. See https://docs.snowflake.com/en/sql-reference/sql/create-security-integration-oauth-external
To test that Snowflake is properly configured to use your OAuth Credentials, you can acquire an access token, and validate it in Snowflake.
Acquire an Access Token using cURL
Example acquiring a token using a client_credentials grant type
curl -X POST -H "Content-Type: application/x-www-form-urlencoded;charset=UTF-8" \
--data-urlencode "client_id=<client_id>" \
--data-urlencode "client_secret=<client_secret>" \
--data-urlencode "grant_type=client_credentials" \
--data-urlencode "scope=<scope>" \
'<access_token_endpoint>'Example acquiring a token using a password grant type
curl -X POST -H "Content-Type: application/x-www-form-urlencoded;charset=UTF-8" \
--data-urlencode "client_id=<client_id>" \
--data-urlencode "client_secret=<client_secret>" \
--data-urlencode "grant_type=password" \
--data-urlencode "username=<username>" \
--data-urlencode "password=<password>" \
--data-urlencode "scope=<scope>" \
'<access_token_endpoint>'Validate Access Token in Snowflake
SELECT SYSTEM$VERIFY_EXTERNAL_OAUTH_TOKEN('<access_token>')Providing service user credentials during onboarding
When using Monte Carlo's onboarding wizard, please provide the following details to validate and complete the integration:
- User: The name of the service user created, MONTE_CARLO by default.
- Private key and optional Private key passphrase for the encrypted private key.
- Account: This would be the part that precedes snowflakecomputing.com in your Snowflake console URL. For example, if you use the URL https://xyz12345.us-east-1.snowflakecomputing.com/ to access Snowflake, your account name will be xyz12345.us-east-1.
- Warehouse: Please provide the name of the warehouse you chose while creating the service user, MONTE_CARLO_WH by default.
Providing credentials with the Terraform provider
The Terraform provider is in beta. It is versioned
0.x, so pin its minor version.
The Terraform provider takes the same details as the wizard. montecarlo_snowflake_credentials holds the user, the private key and its optional passphrase, the account and the warehouse; montecarlo_warehouse and montecarlo_connection put them on a deployment. The private key and passphrase are write-only arguments, so they are never stored in the Terraform state or plan. That needs Terraform 1.11 or later.
resource "montecarlo_warehouse" "snowflake" {
name = "production-snowflake"
type = "snowflake"
deployment_id = var.deployment_id
}
resource "montecarlo_snowflake_credentials" "snowflake" {
account = "xyz12345.us-east-1"
user = "MONTE_CARLO"
private_key_wo = file("${path.module}/snowflake_key.p8")
private_key_wo_version = 1
warehouse = "MONTE_CARLO_WH"
}
resource "montecarlo_connection" "snowflake" {
name = "production-snowflake"
warehouse_id = montecarlo_warehouse.snowflake.id
credentials_id = montecarlo_snowflake_credentials.snowflake.id
}The registry guide Connect a Snowflake warehouse has two complete configurations:
- On a deployment you run: also creates the deployment and deploys and registers a collection agent in your AWS account.
- On the deployment Monte Carlo hosts: Monte Carlo connects from its cloud, with no agent to deploy.
On a deployment you run, the key can stay in your own secret store instead of Monte Carlo's; see Use self-hosted credentials. The provider supports key-pair authentication only; to use External OAuth, connect through the wizard.
Providing credentials with Claude
With the Agent Toolkit, run /monte-carlo-onboarding and Claude does the Monte Carlo steps through the MCP server. The private key never passes through the chat: Claude writes the Terraform or gives you the command, and you run it on your machine. Pick the example closest to your setup:
- Onboard Snowflake with Terraform, key pair stored by Monte Carlo: Monte Carlo connects from its cloud.
- Create the Snowflake service user, agent and connection with Terraform: one apply creates the service user, role, warehouse and grants too, so you skip the setup script.
- Onboard Snowflake through a new AWS agent, all with Terraform: the agent runs in your AWS account and reads the key from your secret store.
- Onboard Postgres and Snowflake with self-hosted credentials: the same without Terraform, using the AWS CLI.
Adding a new query-engine connection
Querying connections enable the creation of custom monitors that leverage multiple different connections. For example, you can now use a larger Snowflake warehouse for resource-intensive metric monitor checks or SQL queries without increasing overall costs or placing additional load on other queries. By default, both the dashboard and APIs will use the primary query engine (i.e., the first query engine added to a warehouse) for all operations.
To add a new querying engine to a Snowflake warehouse, navigate to the Settings > Integrations page, click into the target Snowflake integration, and then click the "Add" button on the right-hand side of the Connections list.

Integration details view, add a new querying engine here
A drawer will open, prompting you to input details for setting up the querying connection. This process is similar to setting up a new connection, so refer to the documentation above for more information.
With the Terraform provider, a querying connection is a second montecarlo_connection on the same warehouse, with its own credentials and job_types limited to the queries it runs:
resource "montecarlo_snowflake_credentials" "query_engine" {
account = "xyz12345.us-east-1"
user = "MONTE_CARLO"
private_key_wo = file("${path.module}/snowflake_key.p8")
private_key_wo_version = 1
warehouse = "MONTE_CARLO_WH_LARGE"
}
resource "montecarlo_connection" "query_engine" {
name = "production-snowflake-large"
warehouse_id = montecarlo_warehouse.snowflake.id
credentials_id = montecarlo_snowflake_credentials.query_engine.id
job_types = ["sql_query", "json_schema"]
depends_on = [montecarlo_connection.snowflake]
}FAQ
Why does the Monte Carlo service user run PUT, GET, and REMOVE commands on @~/mc_stage_check/?
This is Monte Carlo's internal stage access check. Snowflake returns large query results through its internal stage, so if Monte Carlo can't reach the stage, metadata collection can fail even though small test queries succeed. To detect this, Monte Carlo periodically uploads a small test file (a randomly named .bin file containing no customer data) to the service user's own stage (@~/mc_stage_check/), downloads it to a temporary local directory, verifies it, and removes it. The check runs alongside metadata collection and needs no additional grants. If your security tooling alerts on stage GET commands, you can safely allowlist this pattern for the Monte Carlo service user. Blocking it causes the connection validation to fail.
How can I monitor multiple databases with Monte Carlo?
To monitor multiple databases, you will grant Monte Carlo's service user the necessary permissions to each of them. Monte Carlo will automatically monitor all databases to which it has access.
To grant permissions, please run the snippet in the previous section for each of the databases you would like to monitor. You will modify the database_to_monitor parameter in each run.
Can I Remove Permissions from the Monte Carlo User?
If you want to remove permissions from a subset of warehouses, schemas, or tables, run the following commands in the appropriate database, replacing <DATABASE> with the database name and <SCHEMA> with the schema name.
To revoke permissions for a schema:
REVOKE ALL PRIVILEGES ON SCHEMA <DATABASE>.<SCHEMA> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON ALL TABLES IN SCHEMA <DATABASE>.<SCHEMA> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON ALL VIEWS IN SCHEMA <DATABASE>.<SCHEMA> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON FUTURE TABLES IN SCHEMA <DATABASE>.<SCHEMA> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON FUTURE VIEWS IN SCHEMA <DATABASE>.<SCHEMA> FROM MONTE_CARLO_ROLE;As this script removes the USAGE permission from the schema, that's enough to prevent our service user from accessing any objects (tables, views, etc.) in the schema.
To revoke permissions to specific table:
REVOKE ALL PRIVILEGES ON TABLE <DATABASE>.<SCHEMA>.<TABLE> FROM MONTE_CARLO_ROLETo revoke access to the entire database:
REVOKE ALL PRIVILEGES ON DATABASE <DATABASE> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON ALL SCHEMAS IN DATABASE <DATABASE> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON ALL TABLES IN DATABASE <DATABASE> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON ALL VIEWS IN DATABASE <DATABASE> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON FUTURE TABLES IN DATABASE <DATABASE> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON FUTURE VIEWS IN DATABASE <DATABASE> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON FUTURE TABLES IN SCHEMA <DATABASE>.<SCHEMA> FROM MONTE_CARLO_ROLE;
REVOKE ALL PRIVILEGES ON FUTURE VIEWS IN SCHEMA <DATABASE>.<SCHEMA> FROM MONTE_CARLO_ROLE;
MONTE_CARLO_ROLEis usually the role assigned to the user used by Monte Carlo but it might need to be updated too
Can I Update the Service User from Username/Password To Key-Pair Authentication?
Follow these steps if you need to switch your Monte Carlo service user from user/password to key-pair authentication.
- Follow Snowflake's key-pair authentication guide to create a private and public key for the service user to use.
- Generating a private key
- The private key will be needed during onboarding in the Monte Carlo UI.
- Monte Carlo supports using an encrypted or unencrypted private key.
- Generating a public key
- The public key will be assigned to the service user in Snowflake
- Generating a private key
- Using the public key created above, alter the service user in snowflake with this command:
ALTER USER MONTE_CARLO SET TYPE = SERVICE, RSA_PUBLIC_KEY='MIIBIj...';- Edit your Snowflake connection through the UI
Go to Settings -> Integration, find the snowflake connection, click on the three dots per docs here. Under "authentication type", switch the toggle from "Service account" to "Key pair authentication", then upload the private key you created in step 1.
How can I validate Monte Carlo has access to my assets?
After completing the Snowflake integration setup, your Snowflake assets should appear in the Monte Carlo Assets page between a few minutes and one hour. If you don't see your assets after this time period, you can run a validation test to troubleshoot the connection:
- Navigate to Settings → Integrations
- Find your Snowflake integration and click on it
- Select the specific connection you want to test
- Click the Test button from the connection menu
This validation test will help identify any configuration issues that might be preventing your assets from appearing.

When running the validation test, you should see all green checkmarks. If any errors are encountered, the system will provide specific steps to resolve them.
If assets from a specific database or schema are missing you can use the Data Asset validation to test our ability to access them.
Note: Lineage can take up to 24 hours due to batching.
Why am I seeing freshness updates when I only tagged my table or changed its schema?
Monte Carlo retrieves freshness information from Snowflake’s INFORMATION_SCHEMA. Snowflake updates the last altered time of a table when it is tagged, or modified via DDL operations. As a result, Monte Carlo interprets these changes as freshness events, even though the table’s content hasn’t actually changed.
Why does Monte Carlo set STATEMENT_TIMEOUT_IN_SECONDS on its sessions?
So that Snowflake stops a query once Monte Carlo has stopped waiting for it. Otherwise Snowflake would keep running it, up to its own default limit of two days. The value is 850 seconds by default, or a monitor's custom timeout plus 10 seconds.
You can't change the default, but you can choose a longer Timeout, up to 1 hour, on SQL, metric, validation and comparison monitors. If your warehouse has a lower STATEMENT_TIMEOUT_IN_SECONDS, Snowflake applies the lower value.
Details and edge cases
How the Snowflake statement timeout is set. Monte Carlo sets STATEMENT_TIMEOUT_IN_SECONDS 10 seconds above its own time limit, so Snowflake stops the query just after Monte Carlo stops waiting: 850 seconds by default, or 1810 for a monitor with a 1800-second timeout. Sessions for metadata, query log and warehouse usage collection don't set it.
Updated 11 days ago
