* feat: assign partition ranges by rows
* feat: balance partition rows
* feat: get uppoer bound for part nums
* feat: only split in non-compaction seq scan
* fix: parallel scan on multiple sources
* fix: can split check
* feat: scanner prepare by request
* feat: remove scan_parallelism
* docs: upate docs
* chore: update comment
* style: fix clippy
* feat: skip merge and dedup if there is only one source
* chore: Revert "feat: skip merge and dedup if there is only one source"
Since memtable won't do dedup jobs
This reverts commit 2fc7a54b11.
* test: avoid compaction in sqlness window sort test
* chore: do not create semaphore if num partitions is enough
* chore: more assertions
* chore: fix typo
* fix: compaction flag not set
* chore: address review comments
* feat: ttl zero filter
* refactor: use TimeToLive enum
* fix: unit test
* tests: sqlness
* refactor: Option<TTL> None means UNSET
* tests: sqlness
* fix: 10000 years --> forever
* chore: minor refactor from reviews
* chore: rename back TimeToLive
* refactor: split imme request from normal requests
* fix: use correct lifetime
* refactor: rename immediate to instant
* tests: flow sink table default ttl
* refactor: per review
* tests: sqlness
* fix: ttl alter to instant
* tests: sqlness
* refactor: per review
* chore: per review
* feat: add db ttl type&forbid instant for db
* tests: more unit test
* fix: use SchemaCache to locate database metadata
* main:
Refactor SchemaMetadataManager to use TableInfoCacheRef
- Replace TableInfoManagerRef with TableInfoCacheRef in SchemaMetadataManager
- Update DatanodeBuilder to pass TableInfoCacheRef to SchemaMetadataManager
- Rename error MissingCacheRegistrySnafu to MissingCacheSnafu in datanode module
- Adjust tests to use new mock_schema_metadata_manager with TableInfoCacheRef
* fix/schema-cache-invalidation: Add cache module and integrate cache registry into datanode
• Implement build_datanode_cache_registry function to create cache registry for datanode
• Integrate cache registry into datanode by modifying DatanodeBuilder and HeartbeatTask
• Refactor InvalidateTableCacheHandler to InvalidateCacheHandler and move to common-meta crate
• Update Cargo.toml to include cache as a dev-dependency for datanode
• Adjust related modules (flownode, frontend, tests-integration, standalone) to use new cache handler and registry
• Remove obsolete handler module from frontend crate
* fix: fuzz imports
* chore: add some doc for cahce builder functions
* refactor: change table info cache to table schema cache
* fix: remove unused variants
* fix fuzz
* chore: apply suggestion
Co-authored-by: Weny Xu <wenymedia@gmail.com>
* chore: apply suggestion
Co-authored-by: Weny Xu <wenymedia@gmail.com>
* fix: compile
---------
Co-authored-by: dennis zhuang <killme2008@gmail.com>
Co-authored-by: Weny Xu <wenymedia@gmail.com>
* feat: prune in each partition
* chore: change pick log to trace
* chore: add in progress partition scan to metrics
* feat: seqscan support pruning in partition
* chore: remove commented codes
* main:
Add common-meta dependency and implement SchemaMetadataManager
- Introduce `common-meta` as a new dependency in `mito2`.
- Implement `SchemaMetadataManager` for managing schema-level metadata.
- Update `DatanodeBuilder` and `MitoEngine` to pass `KvBackendRef` for schema metadata management.
- Add `SchemaMetadataManager` to `RegionWorkerLoop` for compaction handling.
- Include `SchemaNameKey` usage in compaction-related code.
- Add `database_metadata_manager` module with `SchemaMetadataManager` struct and associated logic.
* fix/database-base-ttl:
Refactor metadata management and update compaction logic
- Remove `database_metadata_manager` and introduce `schema_metadata_manager`
- Update compaction logic to handle TTL based on schema metadata
- Adjust tests to use `schema_metadata_manager` for setting up schema options
- Fix engine creation in tests to pass `kv_backend` explicitly
- Remove unused imports and apply minor code cleanups
* fix/database-base-ttl:
Extend CREATE TABLE LIKE to inherit schema options
- Implement inheritance of database level options for CREATE TABLE LIKE
- Add schema options to SHOW CREATE TABLE output
- Refactor create_table_stmt to include schema_options in SQL generation
- Update error handling to include TableMetadataManagerSnafu
* fix/database-base-ttl:
Refactor error handling and remove schema dependency in table creation
- Replace expect with the ? operator for error handling in open_compaction_region
- Simplify create_logical_tables by removing catalog and schema name parameters
- Remove unnecessary schema retrieval and merging of schema options in create_table_info
- Clean up unused imports and redundant code
* fix/database-base-ttl:
Refactor error handling and update documentation comments
- Update comment to reflect retrieval of schema options instead of metadata
- Introduce new error type `GetSchemaMetadataSnafu` for schema metadata retrieval failures
- Implement error handling for schema metadata retrieval in `find_ttl` function
* fix: toml
* fix/database-base-ttl:
Refactor SchemaMetadataManager and adjust Cargo.toml dependencies
- Remove unused imports in schema_metadata_manager.rs
- Add conditional compilation for SchemaMetadataManager::new
- Update Cargo.toml to remove "testing" feature from common-meta dependency in main section and add it to dev-dependencies
* fix/database-base-ttl:
Fix typos in comments and function names across multiple modules
- Correct spelling of 'parallelism' in region_server, engine, and scan_region modules
- Amend typo in TODO comment from 'persisent' to 'persistent' in server module
- Update incorrect test query from 'versiona' to 'version' in federated module tests
* fix/database-base-ttl: Add schema existence check in StatementExecutor for CREATE TABLE operation
* fix/database-base-ttl: Add warning log for failed TTL retrieval in compaction region open function
* fix/database-base-ttl:
Refactor to use SchemaMetadataManagerRef in Datanode and MitoEngine
- Replace KvBackendRef with SchemaMetadataManagerRef across various components.
- Update DatanodeBuilder and MitoEngine to pass SchemaMetadataManagerRef instead of KvBackendRef.
- Adjust test cases to use get_schema_metadata_manager method for consistency.
fix/database-base-ttl:
Fix typos in comments and function names across multiple modules
- Correct spelling of 'parallelism' in region_server, engine, and scan_region modules
- Amend typo in TODO comment from 'persisent' to 'persistent' in server module
- Update incorrect test query from 'versiona' to 'version' in federated module tests
Co-authored-by: Lei, HUANG <mrsatangel@gmail.com>
* feat: get part range min-max from cache for unordered scan
* feat: seq scan push row groups if num_row_groups > 0
* test: test split
* feat: update comment
* test: fix split test
* refactor: rename get meta data method
* feat: adds index size to region statistics
* feat: adds the number of rows for region statistics
* test: adds sqlness test for region_statistics
* fix: test
* feat/alter-ttl:
Update greptime-proto source and add ChangeTableOptions handling
- Change greptime-proto source repository and revision in Cargo.lock and Cargo.toml
- Implement handling for ChangeTableOptions in grpc-expr and meta modules
- Add support for parsing and applying region option changes in mito2
- Introduce new error type for invalid change table option requests
- Add humantime dependency to store-api
- Fix SQL syntax in tests for changing column types
* chore: remove write buffer size option handling since we don't support specifying write_buffer_size for single table or region
* persist ttl to manifest
* chore: add sqlness
* fix: sqlness
* fix: typo and toml format
* fix: tests
* update: change alter syntax
* feat/alter-ttl: Add Clone trait to RegionFlushRequest and remove redundant Default derive in region_request.rs.
* feat/alter-ttl: Refactor code to replace 'ChangeTableOption' with 'ChangeRegionOption' and handle TTL as a region option
• Rename ChangeTableOption to ChangeRegionOption across various files.
• Update AlterKind::ChangeTableOptions to AlterKind::ChangeRegionOptions.
• Modify TTL handling to treat '0d' as None for TTL in table options.
• Adjust related function names and comments to reflect the change from table to region options.
• Include test case updates to verify the new TTL handling behavior.
* chore: update format
* refactor: update region options in DatanodeTableValue
* feat/alter-ttl:
Remove TTL handling from RegionManifest and related structures
- Eliminate TTL fields from `RegionManifest`, `RegionChange`, and associated handling logic.
- Update tests and checksums to reflect removal of TTL.
- Refactor `RegionOpener` and `handle_alter` to adjust to TTL removal.
- Simplify `RegionChangeResult` by replacing `change` with `new_meta`.
* chore: fmt
* remove useless delete op
* feat/alter-ttl: Updated Cargo.lock and gRPC expression Cargo.toml to include store-api dependency. Refactored alter.rs to use ChangeOption from store-api instead of ChangeTableOptionRequest.
Adjusted error handling in error.rs to use MetadataError. Modified handle_alter.rs to handle TTL changes with ChangeOption. Simplified region_request.rs by replacing
ChangeRegionOption with ChangeOption and removing redundant code. Removed UnsupportedTableOptionChange error in table/src/error.rs. Updated metadata.rs to use ChangeOption for table
options. Removed ChangeTableOptionRequest enum and related conversion code from requests.rs.
* feat/alter-ttl: Update greptime-proto dependency to revision 53ab9a9553
* chore: format code
* chore: update greptime-proto
* feat: add empty batch to end of range stream
* feat: add batch validation
* fix: validate batch order
* fix: not yield empty batch in compaction
* fix: empty record batch
* feat: add a flag to enable empty batch
* feat: define range meta
* feat: group ranges
* feat: split range
* feat: build ranges from the scan input
* feat: get partition range from range meta
* feat: build file range
* feat: unordered scan read by ranges
* feat: wip for mem ranges
* feat: build ranges
* feat: remove unused codes
* chore: update comments
* feat: update metrics
* chore: address review comments
* chore: debug assertion
* Commit Message
Clarify documentation for CompactionOutput struct
Updated the documentation for the `CompactionOutput` struct to specify that the output time range is only relevant for windowed compaction.
* Add max_output_file_size to TwcsPicker and TwcsOptions
- Introduced `max_output_file_size` to `TwcsPicker` struct and its logic to enforce output file size limits during compaction.
- Updated `TwcsOptions` to include `max_output_file_size` and adjusted related tests.
- Modified `new_picker` function to initialize `TwcsPicker` with the new `max_output_file_size` field.
* feat/limit-compaction-output-size:
Refactor compaction picker and TWCS to support append mode and improve options handling
- Update compaction picker to accept a reference to options and append mode flag
- Modify TWCS picker logic to consider append mode when filtering deleted rows
- Remove VersionControl usage in compactor and simplify return type
- Adjust enforce_max_output_size logic in TWCS picker to handle max output file size
- Add append mode flag to TwcsPicker struct
- Fix incorrect condition in TWCS picker for enforcing max output size
- Update region options tests to reflect new max output file size format (1GB and 7MB)
- Simplify InvalidTableOptionSnafu error handling in create_parser
- Add `compaction.twcs.max_output_file_size` to mito engine option keys
* resolve some comments
