mlrun/mlrun v1.0.0 on GitHub

Main features & enhancements

Datastore: Add Snowflake source, @gtopper
Datastore: Add start_time and end_time to BigQuerySource and SnowflakeSource, @gtopper
Datastore: Add Iguazio application partner attribute, @gtopper
Projects: Add/improve methods to the Project class for better usability, @yaronha
Projects: Add tag support to get_artifact_uri(), @yaronha
Runtimes: Add tolerations support, @Tankilevitch
Runtimes: Mask sensitive data, @Hedingber
Runtimes: Add default resources for functions pods, @Tankilevitch
Runtimes: Add flag to allow printing build logs only on failure, @yaronha
Feature Store: Add support for run_id on targets, @benbd86
Feature Store: Support spark dataframe ingestion, @urihoenig
Feature Store: Add option to use online feature service as context manager, @Tankilevitch
Feature Store: Add validation in target path object creation, @benbd86
Feature Store: Performance improvements for preview and hist calculation in Spark engine, @theSaarco
Feature Store: Spark aggregations with support for emit-per-event, @theSaarco
Feature Store: Preserve all columns in Spark aggregation step, @gtopper
Feature Store: Re-enable flush_interval by default, @gtopper
Model Monitoring: Increase batchSize of the stream trigger in the stream function, @katyakats
Pipelines: Enhance kfp pods enrichment with functions and default attributes, @Tankilevitch
Artifacts: Support artifact tags in project APIs, @yaronha
Docs: Doc restructure, @jillnogold
Docs: Improve K8s installation guide, @george0st
Docs: Add doc about spark operator ingestion, @urihoenig
Docs: Enhance get_offline_features()'s with_indexes docstring, @jillnogold
Docs: Add note about column name limitations in feature set aggregations, @jillnogold
MPIJob: Add project secrets to mpijobs, @theSaarco
Dask: Add project secrets to Dask cluster on deployment, @theSaarco
Spark: Support setting different volume mounts for driver and executor and add helper for host path, @Hedingber
Spark: Support auto-add of k8s project secrets, @theSaarco
Builder: Allow to configure whether push/pull registries are insecure, @Hedingber
Builder: Support configuring image name prefix template to enforce, @Hedingber
Builder: Assign more function spec attributes to kaniko pod, @Tankilevitch
DB: Set the default pool size and max overflow to be equal to the API max workers, @Hedingber
API: Fix build status endpoint offset calculation, @Tankilevitch
API: Increase http connection pools maxsize to be equal to number of workers, @Hedingber
API: Set default thread pool max workers to 64, @Hedingber
Nuclio: Set default nuclio runtime to python 3.7 and bump nuclio-jupyter, @Hedingber
Nuclio: Revert default nuclio runtime to python 3.6, @Hedingber
Nuclio: Fix git URL resolution, @yaronha
Nuclio: Fix base image pull secret assignment, @Tankilevitch
Nuclio: Fix runtime handling for Nuclio <1.6.0 (doesn't support python:3.7), @Hedingber
Requirements: Bump nuclio-jupyter to 0.8.23, @Tankilevitch
Requirements: Limit click versions to 8.0.x, @Tankilevitch
Requirements: Downgrade Tensorflow to 2.4.x and Horovod to 0.22.1, @Hedingber
Requirements: Bump storey version to 1.0.3, @benbd86
Requirements: Bump pip to 22.0.x, @Hedingber
UI: Release notes

More info can be found in the RCs release notes: