-
Notifications
You must be signed in to change notification settings - Fork 80
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Ensure proper sequencing of view migrations #1157
Ensure proper sequencing of view migrations #1157
Conversation
…ergnaud/ucx into view-migrations-sequencing
…ergnaud/ucx into view-migrations-sequencing # Conflicts: # tests/unit/hive_metastore/tables/tables_and_views.json # tests/unit/hive_metastore/test_views.py computes valid sequence of direct views (no sequence required)
if table.view_text is None: | ||
self.table_dependencies.add(table) | ||
else: | ||
self.view_dependencies.add(table) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
do we really need both fields? View can depend on both table and a view...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
for sequencing, we're only concerned about the view dependencies, and we don't want to filter the entire list on each sequencing pass
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
and when debugging it's convenient to see both sets, hence why I keep table_dependencies
self.crawler = crawler | ||
self.result_view_list: list[ViewToMigrate] = [] | ||
self.result_tables_set: set[Table] = set() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
self.crawler = crawler | |
self.result_view_list: list[ViewToMigrate] = [] | |
self.result_tables_set: set[Table] = set() | |
self._crawler = crawler | |
self._result_view_list: list[ViewToMigrate] = [] | |
self._result_tables_set: set[Table] = set() |
need to enforce it in the styleguide :)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Done
self.view_dependencies = set() | ||
|
||
def compute_dependencies(self, all_tables: dict[str, Table]): | ||
if len(self.table_dependencies) + len(self.view_dependencies) == 0: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
if len(self.table_dependencies) + len(self.view_dependencies) == 0: | |
if len(self.table_dependencies) + len(self.view_dependencies) > 0: | |
return |
can we invert the condition to reduce nesting?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Done
self.table = table | ||
self.table_dependencies = set() | ||
self.view_dependencies = set() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
self.table = table | |
self.table_dependencies = set() | |
self.view_dependencies = set() | |
self._table = table | |
self._table_dependencies = set() | |
self._view_dependencies = set() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Done
view.compute_dependencies(all_tables) | ||
not_batched_yet = list(filter(lambda v: v not in self.result_tables_set, view.view_dependencies)) | ||
if len(not_batched_yet) == 0: | ||
result.add(view) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
view.compute_dependencies(all_tables) | |
not_batched_yet = list(filter(lambda v: v not in self.result_tables_set, view.view_dependencies)) | |
if len(not_batched_yet) == 0: | |
result.add(view) | |
for dep in view.dependencies(all_tables): | |
if dep in self._result_tables_set: | |
continue | |
result.add(view) |
this should be more readable. please hide compute_dependencies
as private method there.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Are you sure your version is more readable ? It also adds the view more than once, not great...
Check latest version
self.table_dependencies = set() | ||
self.view_dependencies = set() | ||
|
||
def compute_dependencies(self, all_tables: dict[str, Table]): |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
i'd prefer if this method is hidden and the only method we expose is to iterate over dependencies, while we lazily call this behind the cover.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
done
self.result_tables_set: set[Table] = set() | ||
|
||
def sequence(self) -> list[Table]: | ||
# sequencing is achieved using a very simple algorithm: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
that's a neat algorithm. shouln't we integrate it with Index
here to line up only the views with only migrated dependencies?..
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Possibly, but I don't yet know enough details about the migration integration to provide an opinion. Happy to add that suggestion to a ticket.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@nfx , it's not neat, can improve a lot. Please have a look at my below comment.
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #1157 +/- ##
=======================================
Coverage 89.80% 89.80%
=======================================
Files 61 62 +1
Lines 7249 7340 +91
Branches 1300 1318 +18
=======================================
+ Hits 6510 6592 +82
- Misses 475 479 +4
- Partials 264 269 +5 ☔ View full report in Codecov by Sentry. |
2 unit tests fail: |
self.result_tables_set: set[Table] = set() | ||
|
||
def sequence(self) -> list[Table]: | ||
# sequencing is achieved using a very simple algorithm: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@nfx , it's not neat, can improve a lot. Please have a look at my below comment.
Co-authored-by: Ganesh Girase <ganeshgirase@gmail.com>
* Ensure proper sequencing of view migrations ([#1157](#1157)). In this release, we have introduced a `views_migrator` module and corresponding test cases to ensure proper sequencing of view migrations, addressing issue [#1132](#1132). The module contains two main classes: `ViewToMigrate` and `ViewsMigrator`. The former is responsible for parsing a view's SQL text and identifying its dependencies, while the latter sequences views based on their dependencies. The commit also adds a new method, `__hash__`, to the Table class, which returns a hash value of the key of the table, improving the handling of Table objects. Additionally, we have added unit tests and verified the changes on a staging environment. We have also introduced a new file `tables_and_views.json` for unit testing and added a `views_migrator` module that takes a `TablesCrawler` object and returns a sequence of tables (views) that need to be migrated in the correct order. The commit addresses various scenarios such as no views, direct views, indirect views, deep indirect views, invalid SQL, invalid SQL tables, and circular view references. This release is focused on improving the sequencing of view migrations and is accompanied by appropriate tests. * Experimental support for scanning Delta Tables inside Mount Points ([#1095](#1095)). This commit introduces experimental support for scanning Delta Tables located inside mount points using a new `TablesInMounts` crawler. Users can now scan specific mount points using the `--include-mounts` flag and include Parquet files in the scan results with the `--include-parquet-files` flag. Additionally, the `--filter-paths` flag allows for filtering paths in a mount point and the `--max-depth` flag (currently unimplemented) will filter at a specific sub-folder depth in future development. The project dependencies have been updated to use `databricks-labs-lsql~=0.3.0`. This new feature provides a more granular and flexible way to scan Delta Tables, making the project more user-friendly and adaptable to various use cases. * Fixed `NULL` values in `ucx.views.table_format` to have `UNKNOWN` value instead ([#1156](#1156)). This commit includes a fix for handling NULL values in the `table_format` column of Views in the `ucx.views.table_format` module. Previously, NULL values were displayed as-is, but now they will be replaced with the string "UNKNOWN". This change is part of the fix for issue [#115](#115) * Fixing run_workflow functionality for better error handling ([#1159](#1159)). In this release, the `run_workflow` method in the `workflows.py` file has been updated to improve error handling by waiting for the job to terminate or skip before raising an error, allowing for a more detailed error message to be generated. A new method, `job_initial_run`, has been added to initiate a job run and return the run ID, raising a `NotFound` exception if the job run is not found. The `run_workflow` functionality in the `WorkflowsInstall` module has also been enhanced to handle unexpected error types and improve overall error handling during the installation of products. New test cases have been added and existing ones updated to check how the code handles errors when the run ID is not found or when an `OperationFailed` exception is raised during the installation process. These changes improve the robustness and stability of the system. * Use experimental Permissions Migration API also for Legacy Table ACLs ([#1161](#1161)). This release introduces several changes to the group permissions migration functionality and associated tests. The experimental Permissions Migration API is now being utilized for Legacy Table ACLs, which has led to the removal of the verification step from the experimental group migration job. The `TableAclSupport` import and class have been removed, as they are no longer needed. A new `apply_to_renamed_groups` method has been added for production usage, and a `apply_to_groups_with_different_names` method has been added for integration testing, both of which are part of the Permissions Migration API. Additionally, two tests have been added to support the experimental permissions migration for a group with the same name in the workspace and account. The `permission_manager` parameter has been removed from several test functions in the `test_generic.py` file and replaced with the `MigrationState` class, which is used directly with the `WorkspaceClient` object to apply permissions to groups with different names. The `test_some_entitlements` function in the `test_scim.py` file has also been updated to use the `MigratedGroup` class and the `MigrationState` class's `apply_to_groups_with_different_names` method. Finally, new tests for the Permissions Migration API have been added to the `test_tacl.py` file in the `tests/integration/workspace_access` directory to verify the behavior of the Permissions Migration API when migrating different grants.
* Ensure proper sequencing of view migrations ([#1157](#1157)). In this release, we have introduced a `views_migrator` module and corresponding test cases to ensure proper sequencing of view migrations, addressing issue [#1132](#1132). The module contains two main classes: `ViewToMigrate` and `ViewsMigrator`. The former is responsible for parsing a view's SQL text and identifying its dependencies, while the latter sequences views based on their dependencies. The commit also adds a new method, `__hash__`, to the Table class, which returns a hash value of the key of the table, improving the handling of Table objects. Additionally, we have added unit tests and verified the changes on a staging environment. We have also introduced a new file `tables_and_views.json` for unit testing and added a `views_migrator` module that takes a `TablesCrawler` object and returns a sequence of tables (views) that need to be migrated in the correct order. The commit addresses various scenarios such as no views, direct views, indirect views, deep indirect views, invalid SQL, invalid SQL tables, and circular view references. This release is focused on improving the sequencing of view migrations and is accompanied by appropriate tests. * Experimental support for scanning Delta Tables inside Mount Points ([#1095](#1095)). This commit introduces experimental support for scanning Delta Tables located inside mount points using a new `TablesInMounts` crawler. Users can now scan specific mount points using the `--include-mounts` flag and include Parquet files in the scan results with the `--include-parquet-files` flag. Additionally, the `--filter-paths` flag allows for filtering paths in a mount point and the `--max-depth` flag (currently unimplemented) will filter at a specific sub-folder depth in future development. The project dependencies have been updated to use `databricks-labs-lsql~=0.3.0`. This new feature provides a more granular and flexible way to scan Delta Tables, making the project more user-friendly and adaptable to various use cases. * Fixed `NULL` values in `ucx.views.table_format` to have `UNKNOWN` value instead ([#1156](#1156)). This commit includes a fix for handling NULL values in the `table_format` column of Views in the `ucx.views.table_format` module. Previously, NULL values were displayed as-is, but now they will be replaced with the string "UNKNOWN". This change is part of the fix for issue [#115](#115) * Fixing run_workflow functionality for better error handling ([#1159](#1159)). In this release, the `run_workflow` method in the `workflows.py` file has been updated to improve error handling by waiting for the job to terminate or skip before raising an error, allowing for a more detailed error message to be generated. A new method, `job_initial_run`, has been added to initiate a job run and return the run ID, raising a `NotFound` exception if the job run is not found. The `run_workflow` functionality in the `WorkflowsInstall` module has also been enhanced to handle unexpected error types and improve overall error handling during the installation of products. New test cases have been added and existing ones updated to check how the code handles errors when the run ID is not found or when an `OperationFailed` exception is raised during the installation process. These changes improve the robustness and stability of the system. * Use experimental Permissions Migration API also for Legacy Table ACLs ([#1161](#1161)). This release introduces several changes to the group permissions migration functionality and associated tests. The experimental Permissions Migration API is now being utilized for Legacy Table ACLs, which has led to the removal of the verification step from the experimental group migration job. The `TableAclSupport` import and class have been removed, as they are no longer needed. A new `apply_to_renamed_groups` method has been added for production usage, and a `apply_to_groups_with_different_names` method has been added for integration testing, both of which are part of the Permissions Migration API. Additionally, two tests have been added to support the experimental permissions migration for a group with the same name in the workspace and account. The `permission_manager` parameter has been removed from several test functions in the `test_generic.py` file and replaced with the `MigrationState` class, which is used directly with the `WorkspaceClient` object to apply permissions to groups with different names. The `test_some_entitlements` function in the `test_scim.py` file has also been updated to use the `MigratedGroup` class and the `MigrationState` class's `apply_to_groups_with_different_names` method. Finally, new tests for the Permissions Migration API have been added to the `test_tacl.py` file in the `tests/integration/workspace_access` directory to verify the behavior of the Permissions Migration API when migrating different grants.
Changes
add views_migrator and corresponding test cases
Linked issues
Resolves #1132
Functionality
databricks labs ucx ...
...
...
Tests