Added databricks labs ucx cluster-remap command to remap legacy cluster configurations to UC-compatible
#994
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #994 +/- ##
==========================================
+ Coverage 89.01% 89.14% +0.13%
==========================================
Files 54 55 +1
Lines 6652 6732 +80
Branches 1197 1214 +17
==========================================
+ Hits 5921 6001 +80
Misses 481 481
Partials 250 250 ☔ View full report in Codecov by Sentry. |
|
❌ 111/112 passed, 2 flaky, 1 failed, 19 skipped, 1h55m35s total ❌ test_running_real_assessment_job: TimeoutError: timed out after 0:20:00: (21m24.81s)Flaky tests:
Running from acceptance #1689 |
1 task
prajin-29
had a problem deploying
to
account-admin
GitHub Actions
Failure
|
@prajin-29 need to understand how this command can be used - in its current state, users have to specify individual cluster_id to be remapped - this won't be of much use to our users |
prajin-29
had a problem deploying
to
account-admin
GitHub Actions
Failure
So in that case should the command iterates through all the clusters and convert that to UC enabled? Instead of specifying the cluster id. |
nfx
requested changes
Mar 1, 2024
I'm wondering if it is a good choice to directly edit existing clusters, which may break codes and query running there. For example AT&T had a lot of jobs failed just because they switch on UC and we automatically change all new job clusters to be single user. We should be careful here. In reality, many customer create new UC clusters and gradually migrate their codes to these new clusters.
|
prajin-29
had a problem deploying
to
account-admin
GitHub Actions
Failure
prajin-29
had a problem deploying
to
account-admin
GitHub Actions
Failure
prajin-29
had a problem deploying
to
account-admin
GitHub Actions
Failure
dmoore247
suggested changes
Mar 19, 2024
| continue | ||
| access_mode = self._get_access_mode(cluster_details.data_security_mode.name) | ||
| self._installation.save(cluster_details, filename=f'backup/clusters/{cluster_details.cluster_id}.json') | ||
| self._ws.clusters.edit( |
There was a problem hiding this comment.
Is there a way to make this resilient to cluster API changes? The Cluster UI and options are constantly changing, how can this code just focus on the specifics of credentials, spark configs and data security mode and pass through all the other configuration stuff. Will this code break frequently? [I don't know]
nfx
requested changes
Mar 20, 2024
prajin-29
had a problem deploying
to
account-admin
GitHub Actions
Failure
prajin-29
had a problem deploying
to
account-admin
GitHub Actions
Failure
nfx
requested changes
Mar 20, 2024
| for cluster in cluster_list: | ||
| try: | ||
| cluster_details = self._installation.load(ClusterDetails, filename=f"/backup/clusters/{cluster}.json") | ||
| if cluster_details.spark_version is None: |
There was a problem hiding this comment.
spark version may be absent if cluster is using a policy, so this line is not necessary
prajin-29
had a problem deploying
to
account-admin
GitHub Actions
Failure
prajin-29
had a problem deploying
to
account-admin
GitHub Actions
Failure
nfx
approved these changes
Mar 20, 2024
nfx
changed the title
databricks labs ucx cluster-remapdatabricks labs ucx cluster-remap command to remap legacy cluster configurations to UC-compatible
|
failed test run is unrelated to code added. merging. |
nfx
added a commit
that referenced
this pull request
Mar 21, 2024
* Added Legacy Table ACL grants migration ([#1054](#1054)). This commit introduces a legacy table ACL grants migration to the `migrate-tables` workflow, resolving issue [#340](#340) and paving the way for follow-up PRs [#887](#887) and [#907](#907). A new `GrantsCrawler` class is added for crawling grants, along with a `GroupManager` class to manage groups during migration. The `TablesMigrate` class is updated to accept an instance of `GrantsCrawler` and `GroupManager` in its constructor. The migration process has been thoroughly tested with unit tests, integration tests, and manual testing on a staging environment. The changes include the addition of a new Enum class `AclMigrationWhat` and updates to the `Table` dataclass, and affect the way tables are selected for migration based on rules. The logging and error handling have been improved in the `skip_schema` function. * Added `databricks labs ucx cluster-remap` command to remap legacy cluster configurations to UC-compatible ([#994](#994)). In this open-source library update, we have developed and added the `databricks labs ucx cluster-remap` command, which facilitates the remapping of legacy cluster configurations to UC-compatible ones. This new CLI command comes with user documentation to guide the cluster remapping process. Additionally, we have expanded the functionality of creating and managing UC external catalogs and schemas with the inclusion of `create-catalogs-schemas` and `revert-cluster-remap` commands. This change does not modify existing commands or workflows and does not introduce new tables. The `databricks labs ucx cluster-remap` command allows users to re-map and revert the re-mapping of clusters from Unity Catalog (UC) using the CLI, ensuring compatibility and streamlining the migration process. The new command and associated functions have been manually tested for functionality. * Added `migrate-tables` workflow ([#1051](#1051)). The `migrate-tables` workflow has been added, which allows for more fine-grained control over the resources allocated to the workspace. This workflow includes two new instance variables `min_workers` and `max_workers` in the `WorkspaceConfig` class, with default values of 1 and 10 respectively. A new `trigger` function has also been introduced, which initializes a configuration, SQL backend, and WorkspaceClient based on the provided configuration file. The `run_task` function has been added, which looks up the specified task, logs relevant information, and runs the task's function with the provided arguments. The `Task` class's `fn` attribute now includes an `Installation` object as a parameter. Additionally, a new `migrate-tables` workflow has been added for migrating tables from the Hive Metastore to the Unity Catalog, along with new classes and methods for table mapping, migration status refreshing, and migrating tables. The `migrate_dbfs_root_delta_tables` and `migrate_external_tables_sync` methods perform migrations for Delta tables located in the DBFS root and synchronize external tables, respectively. These functions use the workspace client to access the catalogs and ensure proper migration. Integration tests have also been added for these new methods to ensure their correct operation. * Added handling for `SYNC` command failures ([#1073](#1073)). This pull request introduces changes to improve handling of `SYNC` command failures during external table migrations in the Hive metastore. Previously, the `SYNC` command's result was not checked, and failures were not logged. Now, the `_migrate_external_table` method in `table_migrate.py` fetches the result of the `SYNC` command execution, logs a warning message for failures, and returns `False` if the command fails. A new integration test has been added to simulate a failed `SYNC` command due to a non-existent catalog and schema, ensuring the migration tool handles such failures. A new test case has also been added to verify the handling of `SYNC` command failures during external table migrations, using a mock backend to simulate failures and checking for appropriate log messages. These changes enhance the reliability and robustness of the migration process, providing clearer error diagnosis and handling for potential `SYNC` command failures. * Added initial version of `databricks labs ucx migrate-local-code` command ([#1067](#1067)). A new `databricks labs ucx migrate-local-code` command has been added to facilitate migration of local code to a Databricks environment, specifically targeting Python and SQL files. This initial version is experimental and aims to help users and administrators manage code migration, maintain consistency across workspaces, and enhance compatibility with the Unity Catalog, a component of Databricks' data and AI offerings. The command introduces a new `Files` class for applying migrations to code files, considering their language. It also updates the `.gitignore` file and the pyproject.toml file to ensure appropriate version control management. Additionally, new classes and methods have been added to support code analysis, transformation, and linting for various programming languages. These improvements will aid in streamlining the migration process and ensuring compatibility with Databricks' environment. * Added instance pool to cluster policy ([#1078](#1078)). A new field, `instance_pool_id`, has been added to the cluster policy configuration in `policy.py`, allowing users to specify the ID of an instance pool to be applied to all workflow clusters in the policy. This ID can be manually set or automatically retrieved by the system. A new private method, `_get_instance_pool_id()`, has been added to handle the retrieval of the instance pool ID. Additionally, a new test for table migration jobs has been added to `test_installation.py` to ensure the migration job is correctly configured with the specified parallelism, minimum and maximum number of workers, and instance pool ID. A new test case for creating a cluster policy with an instance pool has also been added to `tests/unit/installer/test_policy.py` to ensure the instance pool is added to the cluster policy during creation. These changes provide users with more control over instance pools and cluster policies, and improve the overall functionality of the library. * Fixed `ucx move` logic for `MANAGED` & `EXTERNAL` tables ([#1062](#1062)). The `ucx move` command has been updated to allow for the movement of UC tables/views after the table upgrade process, providing flexibility in managing catalog structure. The command now supports moving multiple tables simultaneously, dropping managed tables/views upon confirmation, and deep-cloning managed tables while dropping and recreating external tables. A refactoring of the `TableMove` class has improved code organization and readability, and the associated unit tests have been updated to reflect these changes. This feature is targeted towards developers and administrators seeking to adjust their catalog structure after table upgrades, with the added ability to manage exceptional conditions gracefully. * Fixed integration testing with random product names ([#1074](#1074)). In the recent update, the `trigger` function in the `tasks.py` module of the `ucx` framework has undergone modification to incorporate a new argument, `install_folder`, within the `Installation` object. This object is now generated locally within the `trigger` function and subsequently passed to the `run_task` function. The `install_folder` is determined by obtaining the parent directory of the `config_path` variable, transforming it into a POSIX-style path, and eliminating the leading "/Workspace" prefix. This enhancement guarantees that the `run_task` function acquires the correct installation folder for the `ucx` framework, thereby improving the overall functionality and precision of the framework. Furthermore, the `Installation.current` method has been supplanted with the newly formed `Installation` object, which now encompasses the `install_folder` argument. * Refactor installer to separate workflows methods from the installer class ([#1055](#1055)). In this release, the installer in the `cli.py` file has been refactored to improve modularity and maintainability. The installation and workflow functionalities have been separated by importing a new class called `WorkflowsInstallation` from `databricks.labs.ucx.installer.workflows`. The `WorkspaceInstallation` class is no longer used in various functions, and the new `WorkflowsInstallation` class is used instead. Additionally, a new mixin class called `InstallationMixin` has been introduced, which includes methods for uninstalling UCX, removing jobs, and validating installation steps. The `WorkflowsInstallation` class now inherits from this mixin class. A new file, `workflows.py`, has been added to the `databricks/labs/ucx/installer` directory, which contains methods for managing Databricks jobs. The new `WorkflowsInstallation` class is responsible for deploying workflows, uploading wheels to DBFS or WSFS, and creating debug notebooks. The refactoring also includes the addition of new methods for handling specific workflows, such as `run_workflow`, `validate_step`, and `repair_run`, which are now contained in the `WorkflowsInstallation` class. The `test_install.py` file in the `tests/unit` directory has also been updated to include new imports and test functions to accommodate these changes. * Skip unsupported locations while migrating to external location in Azure ([#1066](#1066)). In this release, we have updated the functionality of migrating to an external location in Azure. A new private method `_filter_unsupported_location` has been added to the `locations.py` file, which checks if the location URLs are supported and removes the unsupported ones from the list. Only locations starting with "abfss://" are considered supported. Unsupported locations are logged with a warning message. Additionally, a new test `test_skip_unsupported_location` has been introduced to verify that the `location_migration` function correctly skips unsupported locations during migration to external locations in Azure. The test checks if the correct log messages are generated for skipped unsupported locations, and it mocks various scenarios such as crawled HMS external locations, storage credentials, UC external locations, and installation with permission mapping. The mock crawled HMS external locations contain two unsupported locations: `adl://` and `wasbs://`. This ensures that the function handles unsupported locations correctly, avoiding any unnecessary errors or exceptions during migration. * Triggering Assessment Workflow from Installer based on User Prompt ([#1007](#1007)). A new functionality has been added to the installer that allows users to trigger an assessment workflow based on a prompt during the installation process. The `_trigger_workflow` method has been implemented, which can be initiated with a step string argument. This method retrieves the job ID for the specified step from the `_state.jobs` dictionary, generates the job URL, and triggers the job using the `run_now` method from the `jobs` class of the Workspace object. Users will be asked to confirm triggering the assessment workflow and will have the option to open the job URL in a web browser after triggering it. A new unit test, `test_triggering_assessment_wf`, has been introduced to the `test_install.py` file to verify the functionality of triggering an assessment workflow based on user prompt. This test uses existing classes and functions, such as `MockBackend`, `MockPrompts`, `WorkspaceConfig`, and `WorkspaceInstallation`, to run the `WorkspaceInstallation.run` method with a mocked `WorkspaceConfig` object and a mock installation. The test also includes a user prompt to confirm triggering the assessment job and opening the assessment job URL. The new functionality and test improve the installation process by enabling users to easily trigger the assessment workflow based on their specific needs. * Updated README.md for Service Principal Installation Limit ([#1076](#1076)). This release includes an update to the README.md file to clarify that installing UCX with a Service Principal is not supported. Previously, the file indicated that Databricks Workspace Administrator privileges were required for the user running the installation, but did not explicitly state that Service Principal installation is not supported. The updated text now includes this information, ensuring that users have a clear understanding of the requirements and limitations of the installation process. The rest of the file remains unchanged and continues to provide instructions for installing UCX, including required software and network access. No new methods or functionality have been added, and no existing functionality has been changed beyond the addition of this clarification. The changes in this release have been manually tested to ensure they are functioning as intended.
Merged
nfx
added a commit
that referenced
this pull request
Mar 21, 2024
* Added Legacy Table ACL grants migration ([#1054](#1054)). This commit introduces a legacy table ACL grants migration to the `migrate-tables` workflow, resolving issue [#340](#340) and paving the way for follow-up PRs [#887](#887) and [#907](#907). A new `GrantsCrawler` class is added for crawling grants, along with a `GroupManager` class to manage groups during migration. The `TablesMigrate` class is updated to accept an instance of `GrantsCrawler` and `GroupManager` in its constructor. The migration process has been thoroughly tested with unit tests, integration tests, and manual testing on a staging environment. The changes include the addition of a new Enum class `AclMigrationWhat` and updates to the `Table` dataclass, and affect the way tables are selected for migration based on rules. The logging and error handling have been improved in the `skip_schema` function. * Added `databricks labs ucx cluster-remap` command to remap legacy cluster configurations to UC-compatible ([#994](#994)). In this open-source library update, we have developed and added the `databricks labs ucx cluster-remap` command, which facilitates the remapping of legacy cluster configurations to UC-compatible ones. This new CLI command comes with user documentation to guide the cluster remapping process. Additionally, we have expanded the functionality of creating and managing UC external catalogs and schemas with the inclusion of `create-catalogs-schemas` and `revert-cluster-remap` commands. This change does not modify existing commands or workflows and does not introduce new tables. The `databricks labs ucx cluster-remap` command allows users to re-map and revert the re-mapping of clusters from Unity Catalog (UC) using the CLI, ensuring compatibility and streamlining the migration process. The new command and associated functions have been manually tested for functionality. * Added `migrate-tables` workflow ([#1051](#1051)). The `migrate-tables` workflow has been added, which allows for more fine-grained control over the resources allocated to the workspace. This workflow includes two new instance variables `min_workers` and `max_workers` in the `WorkspaceConfig` class, with default values of 1 and 10 respectively. A new `trigger` function has also been introduced, which initializes a configuration, SQL backend, and WorkspaceClient based on the provided configuration file. The `run_task` function has been added, which looks up the specified task, logs relevant information, and runs the task's function with the provided arguments. The `Task` class's `fn` attribute now includes an `Installation` object as a parameter. Additionally, a new `migrate-tables` workflow has been added for migrating tables from the Hive Metastore to the Unity Catalog, along with new classes and methods for table mapping, migration status refreshing, and migrating tables. The `migrate_dbfs_root_delta_tables` and `migrate_external_tables_sync` methods perform migrations for Delta tables located in the DBFS root and synchronize external tables, respectively. These functions use the workspace client to access the catalogs and ensure proper migration. Integration tests have also been added for these new methods to ensure their correct operation. * Added handling for `SYNC` command failures ([#1073](#1073)). This pull request introduces changes to improve handling of `SYNC` command failures during external table migrations in the Hive metastore. Previously, the `SYNC` command's result was not checked, and failures were not logged. Now, the `_migrate_external_table` method in `table_migrate.py` fetches the result of the `SYNC` command execution, logs a warning message for failures, and returns `False` if the command fails. A new integration test has been added to simulate a failed `SYNC` command due to a non-existent catalog and schema, ensuring the migration tool handles such failures. A new test case has also been added to verify the handling of `SYNC` command failures during external table migrations, using a mock backend to simulate failures and checking for appropriate log messages. These changes enhance the reliability and robustness of the migration process, providing clearer error diagnosis and handling for potential `SYNC` command failures. * Added initial version of `databricks labs ucx migrate-local-code` command ([#1067](#1067)). A new `databricks labs ucx migrate-local-code` command has been added to facilitate migration of local code to a Databricks environment, specifically targeting Python and SQL files. This initial version is experimental and aims to help users and administrators manage code migration, maintain consistency across workspaces, and enhance compatibility with the Unity Catalog, a component of Databricks' data and AI offerings. The command introduces a new `Files` class for applying migrations to code files, considering their language. It also updates the `.gitignore` file and the pyproject.toml file to ensure appropriate version control management. Additionally, new classes and methods have been added to support code analysis, transformation, and linting for various programming languages. These improvements will aid in streamlining the migration process and ensuring compatibility with Databricks' environment. * Added instance pool to cluster policy ([#1078](#1078)). A new field, `instance_pool_id`, has been added to the cluster policy configuration in `policy.py`, allowing users to specify the ID of an instance pool to be applied to all workflow clusters in the policy. This ID can be manually set or automatically retrieved by the system. A new private method, `_get_instance_pool_id()`, has been added to handle the retrieval of the instance pool ID. Additionally, a new test for table migration jobs has been added to `test_installation.py` to ensure the migration job is correctly configured with the specified parallelism, minimum and maximum number of workers, and instance pool ID. A new test case for creating a cluster policy with an instance pool has also been added to `tests/unit/installer/test_policy.py` to ensure the instance pool is added to the cluster policy during creation. These changes provide users with more control over instance pools and cluster policies, and improve the overall functionality of the library. * Fixed `ucx move` logic for `MANAGED` & `EXTERNAL` tables ([#1062](#1062)). The `ucx move` command has been updated to allow for the movement of UC tables/views after the table upgrade process, providing flexibility in managing catalog structure. The command now supports moving multiple tables simultaneously, dropping managed tables/views upon confirmation, and deep-cloning managed tables while dropping and recreating external tables. A refactoring of the `TableMove` class has improved code organization and readability, and the associated unit tests have been updated to reflect these changes. This feature is targeted towards developers and administrators seeking to adjust their catalog structure after table upgrades, with the added ability to manage exceptional conditions gracefully. * Fixed integration testing with random product names ([#1074](#1074)). In the recent update, the `trigger` function in the `tasks.py` module of the `ucx` framework has undergone modification to incorporate a new argument, `install_folder`, within the `Installation` object. This object is now generated locally within the `trigger` function and subsequently passed to the `run_task` function. The `install_folder` is determined by obtaining the parent directory of the `config_path` variable, transforming it into a POSIX-style path, and eliminating the leading "/Workspace" prefix. This enhancement guarantees that the `run_task` function acquires the correct installation folder for the `ucx` framework, thereby improving the overall functionality and precision of the framework. Furthermore, the `Installation.current` method has been supplanted with the newly formed `Installation` object, which now encompasses the `install_folder` argument. * Refactor installer to separate workflows methods from the installer class ([#1055](#1055)). In this release, the installer in the `cli.py` file has been refactored to improve modularity and maintainability. The installation and workflow functionalities have been separated by importing a new class called `WorkflowsInstallation` from `databricks.labs.ucx.installer.workflows`. The `WorkspaceInstallation` class is no longer used in various functions, and the new `WorkflowsInstallation` class is used instead. Additionally, a new mixin class called `InstallationMixin` has been introduced, which includes methods for uninstalling UCX, removing jobs, and validating installation steps. The `WorkflowsInstallation` class now inherits from this mixin class. A new file, `workflows.py`, has been added to the `databricks/labs/ucx/installer` directory, which contains methods for managing Databricks jobs. The new `WorkflowsInstallation` class is responsible for deploying workflows, uploading wheels to DBFS or WSFS, and creating debug notebooks. The refactoring also includes the addition of new methods for handling specific workflows, such as `run_workflow`, `validate_step`, and `repair_run`, which are now contained in the `WorkflowsInstallation` class. The `test_install.py` file in the `tests/unit` directory has also been updated to include new imports and test functions to accommodate these changes. * Skip unsupported locations while migrating to external location in Azure ([#1066](#1066)). In this release, we have updated the functionality of migrating to an external location in Azure. A new private method `_filter_unsupported_location` has been added to the `locations.py` file, which checks if the location URLs are supported and removes the unsupported ones from the list. Only locations starting with "abfss://" are considered supported. Unsupported locations are logged with a warning message. Additionally, a new test `test_skip_unsupported_location` has been introduced to verify that the `location_migration` function correctly skips unsupported locations during migration to external locations in Azure. The test checks if the correct log messages are generated for skipped unsupported locations, and it mocks various scenarios such as crawled HMS external locations, storage credentials, UC external locations, and installation with permission mapping. The mock crawled HMS external locations contain two unsupported locations: `adl://` and `wasbs://`. This ensures that the function handles unsupported locations correctly, avoiding any unnecessary errors or exceptions during migration. * Triggering Assessment Workflow from Installer based on User Prompt ([#1007](#1007)). A new functionality has been added to the installer that allows users to trigger an assessment workflow based on a prompt during the installation process. The `_trigger_workflow` method has been implemented, which can be initiated with a step string argument. This method retrieves the job ID for the specified step from the `_state.jobs` dictionary, generates the job URL, and triggers the job using the `run_now` method from the `jobs` class of the Workspace object. Users will be asked to confirm triggering the assessment workflow and will have the option to open the job URL in a web browser after triggering it. A new unit test, `test_triggering_assessment_wf`, has been introduced to the `test_install.py` file to verify the functionality of triggering an assessment workflow based on user prompt. This test uses existing classes and functions, such as `MockBackend`, `MockPrompts`, `WorkspaceConfig`, and `WorkspaceInstallation`, to run the `WorkspaceInstallation.run` method with a mocked `WorkspaceConfig` object and a mock installation. The test also includes a user prompt to confirm triggering the assessment job and opening the assessment job URL. The new functionality and test improve the installation process by enabling users to easily trigger the assessment workflow based on their specific needs. * Updated README.md for Service Principal Installation Limit ([#1076](#1076)). This release includes an update to the README.md file to clarify that installing UCX with a Service Principal is not supported. Previously, the file indicated that Databricks Workspace Administrator privileges were required for the user running the installation, but did not explicitly state that Service Principal installation is not supported. The updated text now includes this information, ensuring that users have a clear understanding of the requirements and limitations of the installation process. The rest of the file remains unchanged and continues to provide instructions for installing UCX, including required software and network access. No new methods or functionality have been added, and no existing functionality has been changed beyond the addition of this clarification. The changes in this release have been manually tested to ensure they are functioning as intended.
dmoore247
pushed a commit
that referenced
this pull request
Mar 23, 2024
…ster configurations to UC-compatible (#994) ## Changes Adding Cli command to Remap the cluster to UC `databricks labs ucx cluster-remap` ### Linked issues #928 Resolves #.. ### Functionality - [ ] added relevant user documentation - [x] added new CLI command - [ ] modified existing command: `databricks labs ucx ...` - [ ] added a new workflow - [ ] modified existing workflow: `...` - [ ] added a new table - [ ] modified existing table: `...` ### Tests <!-- How is this tested? Please see the checklist below and also describe any other relevant tests --> - [ ] manually tested - [ ] added unit tests - [ ] added integration tests - [ ] verified on staging environment (screenshot attached) --------- Co-authored-by: Serge Smertin <259697+nfx@users.noreply.github.com>
dmoore247
pushed a commit
that referenced
this pull request
Mar 23, 2024
* Added Legacy Table ACL grants migration ([#1054](#1054)). This commit introduces a legacy table ACL grants migration to the `migrate-tables` workflow, resolving issue [#340](#340) and paving the way for follow-up PRs [#887](#887) and [#907](#907). A new `GrantsCrawler` class is added for crawling grants, along with a `GroupManager` class to manage groups during migration. The `TablesMigrate` class is updated to accept an instance of `GrantsCrawler` and `GroupManager` in its constructor. The migration process has been thoroughly tested with unit tests, integration tests, and manual testing on a staging environment. The changes include the addition of a new Enum class `AclMigrationWhat` and updates to the `Table` dataclass, and affect the way tables are selected for migration based on rules. The logging and error handling have been improved in the `skip_schema` function. * Added `databricks labs ucx cluster-remap` command to remap legacy cluster configurations to UC-compatible ([#994](#994)). In this open-source library update, we have developed and added the `databricks labs ucx cluster-remap` command, which facilitates the remapping of legacy cluster configurations to UC-compatible ones. This new CLI command comes with user documentation to guide the cluster remapping process. Additionally, we have expanded the functionality of creating and managing UC external catalogs and schemas with the inclusion of `create-catalogs-schemas` and `revert-cluster-remap` commands. This change does not modify existing commands or workflows and does not introduce new tables. The `databricks labs ucx cluster-remap` command allows users to re-map and revert the re-mapping of clusters from Unity Catalog (UC) using the CLI, ensuring compatibility and streamlining the migration process. The new command and associated functions have been manually tested for functionality. * Added `migrate-tables` workflow ([#1051](#1051)). The `migrate-tables` workflow has been added, which allows for more fine-grained control over the resources allocated to the workspace. This workflow includes two new instance variables `min_workers` and `max_workers` in the `WorkspaceConfig` class, with default values of 1 and 10 respectively. A new `trigger` function has also been introduced, which initializes a configuration, SQL backend, and WorkspaceClient based on the provided configuration file. The `run_task` function has been added, which looks up the specified task, logs relevant information, and runs the task's function with the provided arguments. The `Task` class's `fn` attribute now includes an `Installation` object as a parameter. Additionally, a new `migrate-tables` workflow has been added for migrating tables from the Hive Metastore to the Unity Catalog, along with new classes and methods for table mapping, migration status refreshing, and migrating tables. The `migrate_dbfs_root_delta_tables` and `migrate_external_tables_sync` methods perform migrations for Delta tables located in the DBFS root and synchronize external tables, respectively. These functions use the workspace client to access the catalogs and ensure proper migration. Integration tests have also been added for these new methods to ensure their correct operation. * Added handling for `SYNC` command failures ([#1073](#1073)). This pull request introduces changes to improve handling of `SYNC` command failures during external table migrations in the Hive metastore. Previously, the `SYNC` command's result was not checked, and failures were not logged. Now, the `_migrate_external_table` method in `table_migrate.py` fetches the result of the `SYNC` command execution, logs a warning message for failures, and returns `False` if the command fails. A new integration test has been added to simulate a failed `SYNC` command due to a non-existent catalog and schema, ensuring the migration tool handles such failures. A new test case has also been added to verify the handling of `SYNC` command failures during external table migrations, using a mock backend to simulate failures and checking for appropriate log messages. These changes enhance the reliability and robustness of the migration process, providing clearer error diagnosis and handling for potential `SYNC` command failures. * Added initial version of `databricks labs ucx migrate-local-code` command ([#1067](#1067)). A new `databricks labs ucx migrate-local-code` command has been added to facilitate migration of local code to a Databricks environment, specifically targeting Python and SQL files. This initial version is experimental and aims to help users and administrators manage code migration, maintain consistency across workspaces, and enhance compatibility with the Unity Catalog, a component of Databricks' data and AI offerings. The command introduces a new `Files` class for applying migrations to code files, considering their language. It also updates the `.gitignore` file and the pyproject.toml file to ensure appropriate version control management. Additionally, new classes and methods have been added to support code analysis, transformation, and linting for various programming languages. These improvements will aid in streamlining the migration process and ensuring compatibility with Databricks' environment. * Added instance pool to cluster policy ([#1078](#1078)). A new field, `instance_pool_id`, has been added to the cluster policy configuration in `policy.py`, allowing users to specify the ID of an instance pool to be applied to all workflow clusters in the policy. This ID can be manually set or automatically retrieved by the system. A new private method, `_get_instance_pool_id()`, has been added to handle the retrieval of the instance pool ID. Additionally, a new test for table migration jobs has been added to `test_installation.py` to ensure the migration job is correctly configured with the specified parallelism, minimum and maximum number of workers, and instance pool ID. A new test case for creating a cluster policy with an instance pool has also been added to `tests/unit/installer/test_policy.py` to ensure the instance pool is added to the cluster policy during creation. These changes provide users with more control over instance pools and cluster policies, and improve the overall functionality of the library. * Fixed `ucx move` logic for `MANAGED` & `EXTERNAL` tables ([#1062](#1062)). The `ucx move` command has been updated to allow for the movement of UC tables/views after the table upgrade process, providing flexibility in managing catalog structure. The command now supports moving multiple tables simultaneously, dropping managed tables/views upon confirmation, and deep-cloning managed tables while dropping and recreating external tables. A refactoring of the `TableMove` class has improved code organization and readability, and the associated unit tests have been updated to reflect these changes. This feature is targeted towards developers and administrators seeking to adjust their catalog structure after table upgrades, with the added ability to manage exceptional conditions gracefully. * Fixed integration testing with random product names ([#1074](#1074)). In the recent update, the `trigger` function in the `tasks.py` module of the `ucx` framework has undergone modification to incorporate a new argument, `install_folder`, within the `Installation` object. This object is now generated locally within the `trigger` function and subsequently passed to the `run_task` function. The `install_folder` is determined by obtaining the parent directory of the `config_path` variable, transforming it into a POSIX-style path, and eliminating the leading "/Workspace" prefix. This enhancement guarantees that the `run_task` function acquires the correct installation folder for the `ucx` framework, thereby improving the overall functionality and precision of the framework. Furthermore, the `Installation.current` method has been supplanted with the newly formed `Installation` object, which now encompasses the `install_folder` argument. * Refactor installer to separate workflows methods from the installer class ([#1055](#1055)). In this release, the installer in the `cli.py` file has been refactored to improve modularity and maintainability. The installation and workflow functionalities have been separated by importing a new class called `WorkflowsInstallation` from `databricks.labs.ucx.installer.workflows`. The `WorkspaceInstallation` class is no longer used in various functions, and the new `WorkflowsInstallation` class is used instead. Additionally, a new mixin class called `InstallationMixin` has been introduced, which includes methods for uninstalling UCX, removing jobs, and validating installation steps. The `WorkflowsInstallation` class now inherits from this mixin class. A new file, `workflows.py`, has been added to the `databricks/labs/ucx/installer` directory, which contains methods for managing Databricks jobs. The new `WorkflowsInstallation` class is responsible for deploying workflows, uploading wheels to DBFS or WSFS, and creating debug notebooks. The refactoring also includes the addition of new methods for handling specific workflows, such as `run_workflow`, `validate_step`, and `repair_run`, which are now contained in the `WorkflowsInstallation` class. The `test_install.py` file in the `tests/unit` directory has also been updated to include new imports and test functions to accommodate these changes. * Skip unsupported locations while migrating to external location in Azure ([#1066](#1066)). In this release, we have updated the functionality of migrating to an external location in Azure. A new private method `_filter_unsupported_location` has been added to the `locations.py` file, which checks if the location URLs are supported and removes the unsupported ones from the list. Only locations starting with "abfss://" are considered supported. Unsupported locations are logged with a warning message. Additionally, a new test `test_skip_unsupported_location` has been introduced to verify that the `location_migration` function correctly skips unsupported locations during migration to external locations in Azure. The test checks if the correct log messages are generated for skipped unsupported locations, and it mocks various scenarios such as crawled HMS external locations, storage credentials, UC external locations, and installation with permission mapping. The mock crawled HMS external locations contain two unsupported locations: `adl://` and `wasbs://`. This ensures that the function handles unsupported locations correctly, avoiding any unnecessary errors or exceptions during migration. * Triggering Assessment Workflow from Installer based on User Prompt ([#1007](#1007)). A new functionality has been added to the installer that allows users to trigger an assessment workflow based on a prompt during the installation process. The `_trigger_workflow` method has been implemented, which can be initiated with a step string argument. This method retrieves the job ID for the specified step from the `_state.jobs` dictionary, generates the job URL, and triggers the job using the `run_now` method from the `jobs` class of the Workspace object. Users will be asked to confirm triggering the assessment workflow and will have the option to open the job URL in a web browser after triggering it. A new unit test, `test_triggering_assessment_wf`, has been introduced to the `test_install.py` file to verify the functionality of triggering an assessment workflow based on user prompt. This test uses existing classes and functions, such as `MockBackend`, `MockPrompts`, `WorkspaceConfig`, and `WorkspaceInstallation`, to run the `WorkspaceInstallation.run` method with a mocked `WorkspaceConfig` object and a mock installation. The test also includes a user prompt to confirm triggering the assessment job and opening the assessment job URL. The new functionality and test improve the installation process by enabling users to easily trigger the assessment workflow based on their specific needs. * Updated README.md for Service Principal Installation Limit ([#1076](#1076)). This release includes an update to the README.md file to clarify that installing UCX with a Service Principal is not supported. Previously, the file indicated that Databricks Workspace Administrator privileges were required for the user running the installation, but did not explicitly state that Service Principal installation is not supported. The updated text now includes this information, ensuring that users have a clear understanding of the requirements and limitations of the installation process. The rest of the file remains unchanged and continues to provide instructions for installing UCX, including required software and network access. No new methods or functionality have been added, and no existing functionality has been changed beyond the addition of this clarification. The changes in this release have been manually tested to ensure they are functioning as intended.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is the functionality for migrating cluster configurations to UC. The
ClusterAccessclass has several methods:list_clusters: This method lists all the clusters that are not associated with a job in the Databricks workspace._get_access_mode: This method maps data security mode values to theDataSecurityModeenum.map_cluster_to_uc: This method edits a given cluster to be compatible with UC-X by updating its properties such as access mode, spark version, etc. It also saves a backup of the original cluster configuration before editing. If an error occurs during the editing process, the method skips the cluster and logs a warning message.revert_cluster_remap: This method restores the original configuration of a cluster that was previously edited by themap_cluster_to_ucmethod. It loads the backup configuration from a JSON file and applies it to the cluster. If an error occurs during the restoration process, the method skips the cluster and logs a warning message.The
map_cluster_to_ucmethod takes care of editing the cluster configurations whilerevert_cluster_remapcan be used to restore the original configurations if required, making it a useful feature for migrating legacy clusters to the UC-X format.The
ClusterAccessclass uses theWorkspaceClientclass from thedatabricks.sdk.service.computemodule to interact with the Databricks workspace and retrieve information about the clusters. A test suite is added to check the functionality of themap_cluster_to_uc,list_clusters,revert_cluster_remapmethods and also error handling scenarios. This class is the backbone of the new feature which allows to manage and migrate cluster configurations to UC.