November 04-06, 2024 — 8.42 Release Notes

Attention

The release has completed. For differences between the in-advance and final versions of these release notes, see Release notes change log.

New features

Trust Center: Two new scanners in the Security Essentials scanner package

With this release, we are pleased to announce two new Trust Center scanners in the Security Essentials scanner package. This scanner package scans your account to check whether you have set up the following recommendations:

  • You have an authentication policy that enforces all human users to enroll in multi-factor authentication (MFA) if they use passwords to authenticate.

  • You set up an event table if your account enabled event sharing for a native app, so your account receives a copy of the log messages and event information that is shared with the application provider.

For more information, see Security Essentials scanner package.

Serverless alerts — General availability

With this release, we are pleased to announce the general availability of the serverless compute model for Snowflake alerts, which was previously available as a preview feature.

When you configure an alert to use the serverless compute model, Snowflake automatically resizes and scales up or down the compute resources required for the alert. Snowflake determines the ideal size of the compute resources for a given run based on a dynamic analysis of statistics for the most recent previous runs of the same alert.

To use the serverless compute model for an alert, omit the WAREHOUSE parameter when executing the CREATE ALERT command.

For more information, see Setting up alerts based on data in Snowflake.

SQL updates

PARSE_JSON and TRY_PARSE_JSON functions: Duplicate keys are now allowed

With this release, the PARSE_JSON and TRY_PARSE_JSON functions have a new parameter argument. When this argument is set to d, duplicate keys are allowed in the string being parsed. If there are duplicate keys, these functions only return the value associated with the last occurrence of each key.

Extensibility updates

New Tensorflow version might require specifying Keras

With this release, version 2.17.0 of the Tensorflow library has been added. The new Tensorflow version includes a changed module structure for Keras, a deep learning API. If you have user-defined functions (UDFs) or procedures that use Tensorflow but don’t specify a version earlier than 2.17.0, Snowflake will assume that your handler should automatically begin using version 2.17.0 when you execute CREATE OR REPLACE for the UDF or procedure.

When you create or update the UDF or procedure, you might see an error such as the following:

from tensorflow.keras.models import Sequential ModuleNotFoundError: No module named 'tensorflow.keras' in function
Copy

To resolve this error, follow the guidance in Migrating Keras 2 code to multi-backend Keras 3 (https://keras.io/guides/migrating_to_keras_3/). You might need to add Keras as a separate package using the PACKAGES parameter in CREATE OR REPLACE.

Data pipeline updates

Tasks: Serverless tasks user control — General availability

With this release, we are pleased to announce that you can take some control over the cost and performance of serverless tasks by setting the following parameters: SERVERLESS_TASK_MAX_STATEMENT_SIZE, SERVERLESS_TASK_MIN_STATEMENT_SIZE, and TARGET_COMPLETION_INTERVAL.

For more information, see Serverless tasks.

Tasks: Task success notifications — General availability

With this release, we are pleased to announce the general availability of task success notifications. Snowflake can push success notifications to a cloud messaging service when a task graph completes successfully. Success notification integration is only specified on a root task of a task graph. Snowflake only sends success notifications when the entire task graph is successfully executed and will not send notifications for any successfully executed standalone task.

For more information, see Configuring a task to send success notifications.

AI & ML updates

API-level Role-based Access Control (RBAC) for Cortex Analyst

To further enhance security and access management, we are introducing API-level Role-based Access Control (RBAC) for Cortex Analyst. All requests made to Cortex Analyst must use a role which has been granted the CORTEX USER role. This provides admins a way to control who can call Cortex Analyst with Snowflake RBAC. CORTEX_USER is granted to PUBLIC by default. For more information, see Required privileges.

Release notes change log

Announcement

Update

Date

Release notes

Initial publication (preview)

01-Nov-24

API-level Role-based Access Control (RBAC) for Cortex Analyst

Added to AI & ML updates

05-Nov-24

Tasks: Serverless tasks user control

Added to Data pipeline updates section

06-Nov-24

Outbound private connectivity for Snowflake features

Removed from New features section

07-Nov-24

Trust Center: Two new scanners in the Security Essentials scanner package

Added to New features section

11-Nov-24

Language: English