Release Notes July 2026
Orchestrators
Breaking Change: Snowflake Python Connector OCSP Cache Lock Errors in Multiprocessing Jobs
Recent versions of snowflake-connector-python (and its filelock dependency) introduced stricter fork-safety checks on the local OCSP response cache lock file (ocsp_response_validation_cache.json.lock). Python jobs that open a Snowflake connection in the parent process and then fork worker processes with multiprocessing.Pool (the default fork start method on Linux) may now fail with an error similar to:
snowflake.connector.errors.OperationalError: 253003: While putting file(s) there was an error: 'RuntimeError('UnixFileLock on /root/.cache/snowflake/ocsp_response_validation_cache.json.lock was inherited across fork; construct a new instance')'
This can affect any custom Python job that combines multiprocessing with the Snowflake connector, since snowflake-connector-python is typically installed unpinned via requirements.txt and will pick up the latest version on the next image/dependency rebuild.
Workaround: Until a permanent fix is available, create your multiprocessing pool using the spawn start method instead of the platform default (fork):
with multiprocessing.get_context("spawn").Pool(...) as pool:
...
This ensures worker processes start a clean interpreter instead of inheriting the parent process's Snowflake/OCSP state across the fork.
Bug Fix Security Hardening Across All Orchestrators
As part of our ongoing commitment to security, this release resolves a broad set of security findings identified through automated scanning across all orchestrator images. Highlights include:
- Infrastructure as Code (IaC): Remediation of all outstanding IaC scan findings.
- Static Application Security Testing (SAST): Resolution of all outstanding SAST findings.
- Software Composition Analysis (SCA): Remediation of vulnerable and outdated dependencies flagged by SCA scanning.
These changes further strengthen the security posture of the DataOps platform with no impact on existing functionality.
Bug Fix Specified Users/Roles Functionality in SOLE Data Products
Fixed an issue where SOLE Data Products was creating non-interactive users with SYSADMIN grants. The specified users/roles functionality now correctly implements role assignments, preventing unintended privilege escalation.
Feature ROW_TIMESTAMP Parameter Support in Tables
Added support for the ROW_TIMESTAMP parameter in table definitions managed by SOLE. This enables Snowflake's automatic row-level timestamp tracking for tables provisioned through SOLE and the Terraform Provider for Snowflake.
Feature USE_VECTORIZED_SCANNER Parameter Support in File Formats
Added support for the USE_VECTORIZED_SCANNER parameter in file format definitions managed by SOLE. This parameter enables Snowflake's vectorized scanner for improved query performance when reading Parquet files.
Feature Compute Pool Object Support
Added support for Snowflake Compute Pools in SOLE. You can now define and manage compute pools for Snowpark Container Services directly through your SOLE configurations, enabling better resource management for containerized workloads.
Feature Snow CLI and Key Pair Authentication Support in Orchestrators
Enabled Snow CLI and key pair authentication support across multiple orchestrators, including SOLE, Reporting, Snowflake, and VaultSpeed orchestrators.
If you run into any issues with Snow CLI after upgrading, you can temporarily fall back to SnowSQL by setting DATAOPS_USE_SNOWSQL: "true" in your project's variables.yml.
Bug Fix Key Pair Loading Issue in Secrets Manager Orchestrator
Fixed an issue in the Secrets Manager orchestrator where an inherited key pair loading script from the Python orchestrator could cause authentication failures.
Bug Fix Key Pair Authentication Support in Python Orchestrators
Fixed an issue where vault remapping was breaking in Python orchestrators when using key pair authentication. The orchestrators now properly support loading key pair scripts, ensuring secure authentication methods work reliably across all Python-based orchestrators.
Bug Fix dbt 1.11 Manifest Parsing in Compile Models Job
Fixed an issue where dbt-results-parser failed to parse the dbt manifest when using dbt 1.11, causing the Compile Models job in the MATE orchestrator to abort with a Pydantic validation error.
Bug Fix show_initial_rows Parameter for Streams
Fixed an issue with the handling of the show_initial_rows parameter in Stream definitions managed by SOLE.
Bug Fix Default Column Values Conflict in Hybrid Tables
Fixed an issue where default values on columns always caused a conflict in Hybrid Tables managed by SOLE.