ZenML Release Notes

Performance Improvements

This release includes significant performance optimizations for the ZenML server, particularly when handling large-scale deployments:

• Improved database query efficiency: Rewrote filtering queries to eliminate unnecessary sorting during item counting, removed inefficient DISTINCT statements on multiple columns, and optimized OR subqueries for better database performance at scale. PR #4449
• Enhanced API transaction management: Moved cleanup of expired transactions to an independent background thread that runs periodically, significantly improving API response times especially for large payloads like pipeline snapshots with many steps. PR #4453

Logging Enhancements

Logging capabilities have been expanded with new features and improvements:

• Added new create and update endpoints for logs with support for UUIDs in StepRunRequest and PipelineRunRequest
• Introduced workspace ID and name to pipeline run log metadata (with backward compatibility)
• Added zenml.event.type to error messages for better context tracking
• Introduced environment variable to manage maximum log entries per request
• Fixed inconsistent metadata key formatting (standardized zenml. prefix) PR #4405, PR #4467

Dashboard Updates

• Added elapsed time display to step nodes in the DAG visualization for better pipeline monitoring PR #994

Fixed

• Critical data loss bug: Fixed a critical issue in download_artifact_files_from_response that caused silent data corruption when downloading artifacts larger than 8KB. The bug resulted in up to 98%+ data loss for large artifacts by only preserving the last chunk of data. PR #4422
• ZenML Pro migration: Fixed an issue where cookies from local user accounts persisted after migrating a ZenML OSS server to ZenML Pro via organization enrollment, preventing access to migrated resources in the UI. The server now properly rejects these stale cookies. PR #4473
• UV-only environments: Fixed pipeline run crashes in environments using only uv without pip installed. ZenML now falls back to uv pip freeze when pip freeze is unavailable for collecting environment metadata. Also added UV_FREEZE as an export method for Docker builds. PR #4484
• Kubernetes credential expiration: Fixed an issue where Kubernetes credentials issued by service connectors expired while monitoring long-running jobs, causing monitoring failures. Credentials are now properly refreshed during job monitoring. PR #4493
• Improved CLI messaging when attempting to activate a stack without proper permissions

What's Changed

• Add version 0.93.1 to legacy docs by @github-actions[bot] in #4444
• Fix codeql-action paths and Dependabot config by @strickvl in #4435
• Pin qwen-agent<0.0.32 to fix CI import error by @strickvl in #4441
• Add 0.93.2 to the migration tests by @github-actions[bot] in #4446
• Fix data loss in artifact downloads for files larger than 8KB by @anshishrivastava in #4422
• Rewrite the filtering queries to be more efficient by @stefannica in #4449
• Add an example of a hierarchical search agent by @htahir1 in #4427
• Improve API transaction management for better scalability by @stefannica in #4453
• Add zizmor CI workflow for GitHub Actions security scanning by @strickvl in #4451
• Add AGENTS.md for database migrations by @strickvl in #4450
• Updates to the dashboard page by @bcdurak in #4458
• Add Linux Foundation and CNCF Silver Member logos to README by @strickvl in #4460
• Fixes around the logging context and more by @bcdurak in #4405
• Add --output flag to stack and component describe commands by @strickvl in #4454
• Fix the SQL statement used to delete API transactions by @stefannica in #4470
• Reject cookies of local user accounts when enrolled in ZenML Pro by @stefannica in #4473
• Self-hosted ZenML Pro documentation update by @stefannica in #4472
• Fall back to uv pip freeze when pip is missing by @htahir1 in #4484
• Fix duplicate run status (heartbeat response) by @Json-Andriopoulos in #4489
• Add concurrency_policy and starting_deadline_seconds to K8s CronJob config by @strickvl in #4483
• Exclude grpcio 1.68.0-1.70.x in GCP integration to fix shutdown warnings by @strickvl in #4464
• Fix systemic CI flakiness across multiple test clusters by @strickvl in #4471
• Refresh kubernetes credentials while monitoring jobs by @schustmi in #4493
• Logging Metadata Fixes & Additions by @bcdurak in #4467
• Prepare release 0.93.3 by @github-actions[bot] in #4499

New Contributors

• @anshishrivastava made their first contribution in #4422

Full Changelog: 0.93.2...0.93.3

🎨 Dashboard Enhancements

The ZenML Dashboard now provides better visibility into your pipelines and infrastructure:

• Download Pipeline Code: You can now download the code used for a pipeline snapshot directly from the dashboard. A new Download button appears in the "Code Path" section on both the Pipeline Run details page and the Step details sheet, making it easy to retrieve and review the exact code that was executed. PR #4401, PR #989
• Exception Information Display: When dynamic pipeline runs fail, the dashboard now displays detailed exception information, helping you quickly diagnose and troubleshoot issues. PR #4395, PR #990
• Stack & Component Labels: Labels attached to stacks and components are now visible in the dashboard, making it easier to organize and identify your infrastructure resources. PR #992

🔄 Dynamic Pipeline Improvements

Dynamic pipelines are now more robust and easier to work with:

• Proper Environment Configuration: The pipeline environment is now correctly set while running the entrypoint function of dynamic pipelines, ensuring consistent behavior across different execution contexts. PR #4420

🤖 Developer Experience

Claude Code Plugin: A new ZenML Quick Wins skill for Claude Code helps you implement MLOps best practices directly in your AI-assisted coding workflow. The plugin is available through the Claude Code plugin marketplace and includes comprehensive documentation for multiple AI coding tools. PR #4426

Fixed

🚀 Performance & Scalability

• Artifact Download Fix: Resolved an issue where artifact version downloads were failing due to incorrect RBAC checks on the download endpoint. PR #4401

What's Changed

• Add version 0.93.0 to legacy docs by @github-actions[bot] in #4391
• Add 0.93.1 to the migration tests by @github-actions[bot] in #4393
• Add documentation for schedule activate/deactivate and archiving by @strickvl in #4396
• Fix release flow by @schustmi in #4394
• Disable the use of secure cookies with HTTP by @stefannica in #4399
• Fix incorrect claims about ZenML not supporting schedule updates by @strickvl in #4400
• Fix broken Neptune documentation links and add deprecation notice by @strickvl in #4403
• Optimize Claude Code workflow to skip early without @claude mention by @strickvl in #4404
• Much better Pro docs by @AlexejPenner in #4263
• Store ZenML version in environment dict by @schustmi in #4415
• Implement endpoint to download snapshot code by @schustmi in #4401
• DAG step groups by @schustmi in #4375
• Document private secrets behavior in secrets documentation by @strickvl in #4417
• Track exception info for dynamic pipeline runs by @schustmi in #4395
• Set pipeline environment for dynamic pipelines by @schustmi in #4420
• Fix snack-it workflow permission to comment on PRs by @strickvl in #4425
• Fix unknown column warning in schedule list CLI by @strickvl in #4429
• Add ZenML Quick Wins skill for Claude Code by @strickvl in #4426
• Add missing index to the snapshot table by @stefannica in #4431
• Add comprehensive GitHub Actions security hardening with zizmor by @strickvl in #4421
• Faster merging of step and pipeline config by @schustmi in #4434
• Prepare release 0.93.2 by @github-actions[bot] in #4440
• Full Changelog: 0.93.1...0.93.2