apache-airflow-providers-databricks

Changelog¶

6.3.0¶

Features¶

Add cancel_previous_run to DatabricksRunNowOperator (#38702)
add repair_run support to DatabricksRunNowOperator in deferrable mode (#38619)
Adds job_id as path param in update permission (#38962)

Bug Fixes¶

Fix remaining D401 checks (#37434)
Update ACL during job reset (#38741)
Remove extra slash from update permission endpoint (#38918)
DatabricksRunNowOperator: fix typo in latest_repair_id (#39050)

Misc¶

refactor(databricks): remove redundant else block (#38397)
Rename 'DatabricksSqlOperator''s fields' names to comply with templated fields validation (#38052)

6.2.0¶

Features¶

Update DatabricksSqlOperator to work with namedtuple (#37025)

Misc¶

Bump aiohttp min version to avoid CVE-2024-23829 and CVE-2024-23334 (#37110)
feat: Switch all class, functions, methods deprecations to decorators (#36876)
Get rid of pytest-httpx as dependency (#37334)

6.1.0¶

Features¶

[FEAT] adds repair run functionality for databricks (#36601)

Bug Fixes¶

Fix databricks_sql hook query failing on empty result for return_tuple (#36827)
Rename columns to valid namedtuple attributes + ensure Row.fields are retrieved as tuple (#36949)
check status before DatabricksSubmitRunOperator & DatabricksSubmitRunOperator executes in deferrable mode (#36862)

6.0.0¶

Note

This release of provider is only available for Airflow 2.6+ as explained in the Apache Airflow providers support policy.

Breaking changes¶

Return common data structure in DBApi derived classes (#36205)

Bug Fixes¶

Fix: Implement support for 'fetchone()' in the ODBCHook and the Databricks SQL Hook (#36161)

Misc¶

Bump minimum Airflow version in providers to Airflow 2.6.0 (#36017)
fix typos in DatabricksSubmitRunOperator (#36248)
Add code snippet formatting in docstrings via Ruff (#36262)

5.0.1 (YANKED)¶

Warning

This release has been yanked with a reason: The provider DBApiHook output returned broken output.

Misc¶

Make pyodbc.Row and databricks.Row JSON-serializable via new 'make_serializable' method (#32319)

5.0.0¶

Breaking changes¶

The offset parameter has been deprecated from list_jobs in favor of faster pagination with page_token similarly to Databricks API.

Remove offset-based pagination from 'list_jobs' function in 'DatabricksHook' (#34926)

4.7.0¶

Features¶

Add operator to create jobs in Databricks (#35156)

4.6.0¶

Note

This release of provider is only available for Airflow 2.5+ as explained in the Apache Airflow providers support policy.

Features¶

Add 'DatabricksHook' ClusterState (#34643)

Bug Fixes¶

Respect 'soft_fail' parameter in 'DatabricksSqlSensor' (#34544)
Respect 'soft_fail' argument when running DatabricksPartitionSensor (#34517)
Decode response in f-string (#34518)

Misc¶

Bump min airflow version of providers (#34728)
Use aiohttp.BasicAuth instead of HTTPBasicAuth for aiohttp session in databricks hook (#34590)
Update 'list_jobs' function in 'DatabricksHook' to token-based pagination (#33472)

4.5.0¶

Features¶

Add "QUEUED" to RUN_LIFE_CYCLE_STATES following deployement of … (#33886)
allow DatabricksSubmitRunOperator to accept a pipeline name for a pipeline_task (#32903)

Misc¶

Replace sequence concatenation by unpacking in Airflow providers (#33933)
Improve modules import in Airflow providers by some of them into a type-checking block (#33754)
Use literal dict instead of calling dict() in providers (#33761)
Use f-string instead of in Airflow providers (#33752)

4.4.0¶

Note

This release excluded databricks-sql-connector version 2.9.0 due to a bug that it does not properly declare urllib3 for more information please see https://github.com/databricks/databricks-sql-python/issues/190

Features¶

Add Service Principal OAuth for Databricks. (#33005)

Misc¶

Update docs in databricks.py - we use 2.1 now (#32340)
Do not create lists we don't need (#33519)
Refactor: Improve detection of duplicates and list sorting (#33675)
Simplify conditions on len() in other providers (#33569)
Refactor: Simplify code in smaller providers (#33234)

4.3.3¶

Misc¶

Add a new parameter to SQL operators to specify conn id field (#30784)

4.3.2¶

Bug Fixes¶

fix(providers/databricks): remove the execute method from to-be-deprecated DatabricksRunNowDeferrableOperator (#32806)

Misc¶

Add missing execute_complete method for 'DatabricksRunNowOperator' (#32689)
Add more accurate typing for DbApiHook.run method (#31846)

4.3.1¶

Bug Fixes¶

Modify 'template_fields' of 'DatabricksSqlOperator' to support parent class fields (#32253)

Misc¶

Add default_deferrable config (#31712)

4.3.0¶

Note

This release dropped support for Python 3.7

Features¶

add a return when the event is yielded in a loop to stop the execution (#31985)

Bug Fixes¶

Fix type annotation (#31888)
Fix Databricks SQL operator serialization (#31780)
Making Databricks run related multi-query string in one session again (#31898) (#31899)

Misc¶

Remove return statement after yield from triggers class (#31703)
Remove Python 3.7 support (#30963)

4.2.0¶

Note

This release of provider is only available for Airflow 2.4+ as explained in the Apache Airflow providers support policy.

Features¶

Add conditional output processing in SQL operators (#31136)
Add cancel all runs functionality to Databricks hook (#31038)
Add retry param in databrics async operator (#30744)
Add repair job functionality to databricks hook (#30786)
Add 'DatabricksPartitionSensor' (#30980)

Misc¶

Bump minimum Airflow version in providers (#30917)
Deprecate databricks async operator (#30761)

4.1.0¶

Features¶

Add delete inactive run functionality to databricks provider (#30646)
Databricks SQL sensor (#30477)

4.0.1¶

Bug Fixes¶

DatabricksSubmitRunOperator to support taskflow (#29840)

4.0.0¶

Breaking changes¶

The DatabricksSqlHook is now conforming to the same semantics as all the other DBApiHook implementations and returns the same kind of response in its run method. Previously (pre 4.* versions of the provider, the Hook returned Tuple of ("cursor description", "results") which was not compatible with other DBApiHooks that return just "results". After this change (and dependency on common.sql >= 1.3.1), The DatabricksSqlHook returns now "results" only. The description can be retrieved via last_description field of the hook after run method completes.

That makes the DatabricksSqlHook suitable for generic SQL operator and detailed lineage analysis.

If you had custom hooks or used the Hook in your TaskFlow code or custom operators that relied on this behaviour, you need to adapt your DAGs.

The Databricks DatabricksSQLOperator is also more standard and derives from common SQLExecuteQueryOperator and uses more consistent approach to process output when SQL queries are run. However in this case the result returned by execute method is unchanged (it still returns Tuple of ("description", "results") and this Tuple is pushed to XCom, so your DAGs relying on this behaviour should continue working without any change.

Fix errors in Databricks SQL operator introduced when refactoring (#27854)
Bump common.sql provider to 1.3.1 (#27888)

Bug Fixes¶

Fix templating fields and do_xcom_push in DatabricksSQLOperator (#27868)
Fixing the behaviours of SQL Hooks and Operators finally (#27912)

3.4.0 (YANKED)¶

Warning

This release has been yanked with a reason: There is a bug in DatabricsksSQLOperator

Note

This release of provider is only available for Airflow 2.3+ as explained in the Apache Airflow providers support policy.

Misc¶

Move min airflow version to 2.3.0 for all providers (#27196)
Replace urlparse with urlsplit (#27389)

Features¶

Add SQLExecuteQueryOperator (#25717)
Use new job search API for triggering Databricks job by name (#27446)

3.3.0¶

Features¶

DatabricksSubmitRunOperator dbt task support (#25623)

Misc¶

Add common-sql lower bound for common-sql (#25789)
Remove duplicated connection-type within the provider (#26628)

Bug Fixes¶

Databricks: fix provider name in the User-Agent string (#25873)

3.2.0¶

Features¶

Databricks: update user-agent string (#25578)
More improvements in the Databricks operators (#25260)
Improved telemetry for Databricks provider (#25115)
Unify DbApiHook.run() method with the methods which override it (#23971)

Bug Fixes¶

Databricks: fix test_connection implementation (#25114)
Do not convert boolean values to string in deep_string_coerce function (#25394)
Correctly handle output of the failed tasks (#25427)
Databricks: Fix provider for Airflow 2.2.x (#25674)

3.1.0¶

Features¶

Added databricks_conn_id as templated field (#24945)
Add 'test_connection' method to Databricks hook (#24617)
Move all SQL classes to common-sql provider (#24836)

Bug Fixes¶

Update providers to use functools compat for ''cached_property'' (#24582)

3.0.0¶

Breaking changes¶

Note

This release of provider is only available for Airflow 2.2+ as explained in the Apache Airflow providers support policy.

Features¶

Add Deferrable Databricks operators (#19736)
Add git_source to DatabricksSubmitRunOperator (#23620)

Bug Fixes¶

fix: DatabricksSubmitRunOperator and DatabricksRunNowOperator cannot define .json as template_ext (#23622) (#23641)
Fix UnboundLocalError when sql is empty list in DatabricksSqlHook (#23815)

2.7.0¶

Features¶

Update to the released version of DBSQL connector
DatabricksSqlOperator - switch to databricks-sql-connector 2.x
Further improvement of Databricks Jobs operators (#23199)

2.6.0¶

Features¶

More operators for Databricks Repos (#22422)
Add a link to Databricks Job Run (#22541)
Databricks SQL operators are now Python 3.10 compatible (#22886)

Bug Fixes¶

Databricks: Correctly handle HTTP exception (#22885)

Misc¶

Refactor 'DatabricksJobRunLink' to not create ad hoc TaskInstances (#22571)

2.5.0¶

Features¶

Operator for updating Databricks Repos (#22278)

Bug Fixes¶

Fix mistakenly added install_requires for all providers (#22382)

2.4.0¶

Features¶

Add new options to DatabricksCopyIntoOperator (#22076)
Databricks hook - retry on HTTP Status 429 as well (#21852)

Misc¶

Skip some tests for Databricks from running on Python 3.10 (#22221)

2.3.0¶

Features¶

Add-showing-runtime-error-feature-to-DatabricksSubmitRunOperator (#21709)
Databricks: add support for triggering jobs by name (#21663)
Added template_ext = ('.json') to databricks operators #18925 (#21530)
Databricks SQL operators (#21363)

Bug Fixes¶

Fixed changelog for January 2022 (delayed) provider's release (#21439)

Misc¶

Support for Python 3.10
Updated Databricks docs for correct jobs 2.1 API and links (#21494)

2.2.0¶

Features¶

Add 'wait_for_termination' argument for Databricks Operators (#20536)
Update connection object to ''cached_property'' in ''DatabricksHook'' (#20526)
Remove 'host' as an instance attr in 'DatabricksHook' (#20540)
Databricks: fix verification of Managed Identity (#20550)

2.1.0¶

Features¶

Databricks: add more methods to represent run state information (#19723)
Databricks - allow Azure SP authentication on other Azure clouds (#19722)
Databricks: allow to specify PAT in Password field (#19585)
Databricks jobs 2.1 (#19544)
Update Databricks API from 2.0 to 2.1 (#19412)
Authentication with AAD tokens in Databricks provider (#19335)
Update Databricks operators to match latest version of API 2.0 (#19443)
Remove db call from DatabricksHook.__init__() (#20180)

Bug Fixes¶

Fixup string concatenations (#19099)
Databricks hook: fix expiration time check (#20036)

2.0.2¶

Bug Fixes¶

Move DB call out of DatabricksHook.__init__ (#18339)

2.0.1¶

Misc¶

Optimise connection importing for Airflow 2.2.0

2.0.0¶

Breaking changes¶

Auto-apply apply_default decorator (#15667)

Warning

Due to apply_default decorator removal, this version of the provider requires Airflow 2.1.0+. If your Airflow version is < 2.1.0, and you want to install this provider version, first upgrade Airflow to at least version 2.1.0. Otherwise your Airflow package version will be upgraded automatically and you will have to manually run airflow upgrade db to complete the migration.

1.0.1¶

Updated documentation and readme files.

1.0.0¶

Initial version of the provider.