Skip to content

Latest commit

 

History

History
231 lines (140 loc) · 8.56 KB

CHANGELOG.md

File metadata and controls

231 lines (140 loc) · 8.56 KB
Raw

Changelog

Unreleased

  • Updated dependencies, bumped kedro to 0.18.9, mlflow to 2.3.2

0.7.4 - 2023-02-27

  • Removed field validation from resources configuration field - now it can take any custom parameters such as "nvidia.com/gpu":1
  • Added support for kedro namespaces for parameters

0.7.3 - 2022-09-23

  • Fixed plugin config provider so it respects environment provided by the user

0.7.2 - 2022-09-22

  • Fixed compatibility with kedro-mlflow >= 0.8

0.7.1 - 2022-09-21

  • Support latest changes in Dex login screens
  • Improvements in build process (migrated build tool to poetry, quality gate switched to sonarcloud)

0.7.0 - 2022-08-22

  • Added testing and support for python versions 3.9, 3.10
  • Added templating capabilities to docs generator and used them in the docs for kedro versioning
  • Added pre-commit hook for pyspelling check
  • Changed sphinx markdown engine to myst_parser
  • Added CI for spellchecking the documentation with configuration for myst
  • Updated documentation quickstart to workaround known issues and make it work on local kind cluster
  • Updated documentation - added contributing guidelines and setup tips
  • Merged e2e and unittest workflows. Now e2e runs only if unittests succeed.
  • Changed default resource limits in kubeflow.yaml config
  • Added --wait-for-completion and --timeout for kedro kubeflow run-once command
  • Added e2e tests github action for pull requests with kubeflow setup in gcp
  • Added support for extra volumes per node
  • Refactored configuration classes to Pydantic
  • Add support for kedro>=0.18.1,<0.19

0.6.4 - 2022-06-01

  • Added support for specifying tolerations

0.6.3 - 2022-05-10

  • KFP SDK version bumped to 1.8.11 in order to fix misbehaving TTL issue
  • Dropped support for VertexAI, please use kedro-vertexi instead
  • Add Kedro environment name to the pipeline name during upload

0.6.2 - 2022-03-10

  • Added support for defining retry policy for the Kubeflow Pipelines nodes

0.6.1 - 2022-03-07

  • Fixed support for parameters of type datetime.date

0.6.0 - 2022-02-18

  • Kedro pipeline name is now added into Kubeflow pipeline name during upload
  • Project hook that injected environmental variables values into all the configuration files is dropped, with backward compatibility to support these in kubeflow.yaml
  • Added missing on-exit-handler for node_merge_strategy: full
  • Handle KEDRO_ENV enviroment variable

0.5.1 - 2022-01-28

  • Possibility to run custom Kedro pipeline as on-exit-handler

0.5.0 - 2022-01-27

  • Kedro paramters of complex types (lists and dicts) are now supported
  • run_once and schedule accepts Kedro parameters override
  • Names of the one-off runs and scheduled runs are templated with parameters

0.4.8 - 2022-01-10

0.4.7 - 2022-01-05

  • Add kubeflow_run_id tag to MLFlow run when full node merge strategy is used

0.4.6 - 2021-12-23

  • Passing all KEDRO_CONFIG_ environment variables to the pipeline nodes

0.4.5 - 2021-12-22

  • Add node_merge_strategy alongside with full option to run a whole pipeline in one pod

0.4.4 - 2021-09-29

  • Custom networking setup for Vertex AI pipelines run

0.4.3 - 2021-09-27

  • Kedro environment used by kedro kubeflow invocation is passed to the steps
  • A flag to skip steps output artifacts registration in Kubeflow Metadata

0.4.2 - 2021-08-19

  • Improved Vertex scheduling: removal of stale schedules

0.4.1 - 2021-08-18

  • Passing Kedro environment and pipeline name in Vertex nodes
  • Setting artifact type based on catalog layer in Vertex pipeline
  • Added pipeline param to schedule in Vertex

0.4.0 - 2021-08-11

  • Support for kedro-mlflow>=0.7
  • Support of Google Vertex AI Pipelines (EXPERIMENTAL)
  • Ability to access KFP API behind Dex+authservice authentication
  • Support for multi-user KFP setup (experiment namespace passed via run-once or schedule)
  • New config param: max_cache_staleness to avoid caching KFP steps if required

0.3.1 - 2021-05-25

Fixed

  • Prevent KeyError when catalog had entries without filepath.

0.3.0 - 2021-01-29

Added

  • Support to inject Kedro pipeline parameters for the run
  • Ability to specify resources allocation for the nodes
  • Possibility to configure the pipeline description in the config file
  • The registered output artifacts are not exposed in the Pipelines UI
  • Ability to set ttl of the workflow (how long the pods and volumes stay in the system after finish)
  • Removing the inter-steps data volume during workflow removal (with option to disable the removal in the config using flag keep)

0.2.0 - 2021-01-18

Added

  • Ability to change the effective user id for steps if the ownership of the volume needs it
  • Hook that enables TemplatedConfigLoader that supports dynamic config files. Any env variable named KEDRO_CONFIG_<NAME> can be referenced in configuration files as ${name}
  • Added IAP authentication support for MLflow
  • Increased test coverage for the CLI
  • Creating github actions template with kedro kubeflow init --with-github-actions

Fixed

  • Fixed broken kubeflow init command (#29)

0.1.10 - 2021-01-11

Added

  • Added sample support for TemplatedConfigLoader
  • MLFlow support updated to not use nested runs.
  • Simple configuration validation added.
  • Volume init step is now optional (useful, if there is raw data in the image)

0.1.9 - 2021-01-08

Added

  • Support for MLFlow - if the package is installed then additional step is added with parent run init. Then all separate nodes runs register under this run.
  • Support for inter-steps volume: setup (one volume per pipeline run), initial load (the content of data/ directory from the image and mount to all the steps for artifacts passing.
  • kubeflow init command added to generate sample configuration file.

0.1.8 - 2021-01-05

Added

  • Initial release of kedro-kubeflow plugin
  • Ability to run an anonymous pipeline once as within a specified experiment kedro kubeflow run-once.
  • Ability to upload pipeline kedro kubeflow upload-pipeline.
  • Method to schedule runs for most recent version of given pipeline kedro kubeflow schedule
  • Shortcut to open UI for pipelines using kedro kubeflow ui