v.22.5
Bug Fix
56- Extract Version ID from URI and Configure AWS HTTP Request
- Fix opentelemetry_span_log attribute values alias
- Fix Nullable(String) to Nullable(Bool/IPv4/IPv6) Conversion
- Experimental Fix for Mutation Execution in Tables with Object Type Columns
- Kafka Producer Ignores group.id Configuration Warning
- Experimental Update to max_fired_watermark in WindowView Feature
- Fix "Cannot create column of type Set" error for distributed queries with LIMIT BY
- Experimental Update: WindowView WATCH EVENTS Query Enhanced Stability
- Enable enable_global_with_statement for Subqueries
- Fix implicit cast in optimize_skip_unused_shards_rewrite_in
- ILIKE Function on FixedString Columns Could Return Incorrect Results
- Fix GROUP BY with AggregateFunction type
- Experimental Fix for optimize_aggregation_in_order with GROUP BY and Array Functions
- Fixed Performance Degradation in INSERT SELECT Queries
- Experimental Fix for In-Order GROUP BY with Array Aggregate Functions
- Fix Invalid Output in LowCardinality->ArrowDictionary with Non-UInt8 Indexes
- Fixed Inf Issue in quantileTDigest
- Fix External Tables Data Sending in HedgedConnections with Multiple Parallel Replicas
- Fixed TRUNCATE Query Logical Error in Replicated Database
- Experimental Fix for Stuck Issue When Dropping Source Table in WindowView
- Experimental feature: Fix issue #36671 with rocksdb cache
- Experimental feature: Bug fixes for multiple columns in WindowView
- Fix ClickHouse Keeper Bug Causing Corrupted Compressed Log Files
- Fix constant aggregation query result issue #36728
- Experimental Fix for current_size Count in Cache
- Experimental feature: Fix fire in window view with hop window
- Experimental feature: Fix cast in remote fs cached buffer
- Fix Table Creation with flatten_nested = 0 for Nested Columns
- Fix async read issues from remote filesystem with low cardinality
- Experimental Fix for Insertion to Object Type Columns from Multiple Files in ClickHouse
- Fix timeouts in Hedged requests to prevent connection hangs
- Experimental Fix for Distributed Table Bug in groupBitmap Operations
- Experimental Feature: Cache Class Initialization Issue in ClickHouse Test
- Fix Vertical Merges in Wide Parts to Prevent Exceptions
- Fix server reload on port change without waiting for current connections
- Experimental Feature: Ongoing Timeout and Deadlock Issues in Testing
- Experimental feature: Resolve server restart issue with cache configuration changes
- Fix heap-use-after-free in schema inference
- Fixed Query Settings Parsing in CREATE Query Without Engine Specified
- Experimental Fix for Merging Wide Object Types
- Fix format crash with default expressions following EPHEMERAL
- Fix Missing Column Exception with INTERPOLATE in MergeTree Table
- Fix Error with Literals in WHERE Clause for Join Queries
- Fix offset update in ReadBufferFromEncryptedFile to prevent undefined behaviour
- Fix Hostname Checks for Keeper Cluster Configuration
- Fix GROUP BY to Support Executable User Defined Functions
- Fix Exception for Unknown Server Packet in Client
- Experimental: Add Missing Enum Values in system.session_log Table
- Fix S3Cluster Schema Inference Bug Leading to Incomplete Data Retrieval
- Fix nullptr dereference in JOIN and COLUMNS matcher
- Fix ClickHouseDictionarySource Reload for Scalar Subqueries
- Fix Assertion in JOIN Related to Issue #36199
- Fix parsing error for queries with aliases in special operators
- Experimental Fix for Inserting Nested JSON Arrays into Object Columns
- Fix ALTER DROP COLUMN for Nested Columns in ClickHouse
- Fix substring function range error with negative offset and length parameters
Build/Testing/Packaging Improvement
6- ClickHouse Keeper for x86_64 Now Statically Linked with Musl, No System Library Dependencies
- ClickHouse Now Supports PowerPC64LE Architecture with New Builds
- Limit PowerPC Code Generation to Power8 for Compatibility
- Simplify Performance Test for Increased Usability
- Fail Performance Comparison on Report Errors
- Add ZSTD Support for Arrow
Experimental Feature
7- Improve WATCH query latency and cancel operation in WindowView
- Introspection to Remove Filesystem Cache
- Added wyHash64 Hash Function for SQL
- Improvement: SYSTEM SYNC DATABASE REPLICA Query for Replicated Databases
- Improvement of Remote Filesystem Cache and SYSTEM DROP FILESYSTEM CACHE Enhancements
- Improvement to Cast Semistructured Data Columns to Nullable Type
- Improved Communication for Localhost Replica in Parallel Queries
Improvement
39- Show Erroneous File Names in Parsing Errors for Table Functions
- Allowed Runtime Increase of Background Operation Threads in Config
- Now date time conversion functions saturate to zero for pre-1970 timezones
- Add Warning for ClickHouse Log Level "Test" Due to Severe Performance Issues
- Parse Collations in CREATE TABLE: Exception Handling or Ignoring
- Option compatibility_ignore_auto_increment_in_create_table Simplifies MySQL Migration
- Add JSONLines and NDJSON aliases for JSONEachRow
- Limit Max Partitions Queryable per Hive Table to Prevent Resource Overruns
- Added Implicit Cast for h3kRing Function's Second Argument
- Fix Progress Indication for INSERT SELECT in clickhouse-local
- Fix Bug in MergeTree Table Engines for Outdated Parts Removal After Filesystem Failures
- Implemented New Row Policy Handling Mode in Main Configuration
- Play UI: Right-align Nullable Numbers in Table Cells
- Play UI: Display Single Row Results Vertically with Multiple Columns
- Cleanup CSS for Improved Usability in Play UI
- Finalize Write Buffers on Exception to Resolve #36907
- Fix Exception for Obsolete Setting in ClickHouse Configuration
- Add Extra Diagnostic Info for Exception Handling in ClickHouse
- Allow hash functions for Array of Tuples in ClickHouse
- Added user_defined_path config setting
- Allow cluster macro in s3Cluster table function
- Proper cancellation of INSERT queries in clickhouse-client and clickhouse-local
- Allow Query Cancellation in MySQLHandler with Consistent Query ID
- Add is_all_data_sent Column to system.processes with Testing Enhancements
- Metrics for Time Spent Reading from S3 Calculated Correctly
- Allow File Descriptors in ClickHouse Local Table Function
- Allow Tuple Element Names to Start with Digits
- Clickhouse-benchmark Supports Environment Variable Authentication
- clickhouse-keeper: Add Force Recovery Support for Cluster Reconfiguration without Quorum
- Improve JSON Schema Inference
- Refactor Schema Inference Logic with Improved Error Handling
- Add CLUSTER Grant and Backward Compatibility Directive
- Improved Query Execution with Memory Availability Check
- Nullables Detection in Protobuf and ClickHouse Integration Proposal
- Added content_type support in HTTP handler configuration
- Warn for clickhouse-client --file without --external usage
- Improve MySQL Database Engine for Binary(0) DataType Compatibility
- Improve JSON Report for ClickHouse Benchmark
- Server startup issue with external ClickHouse dictionary resolved
New Feature
9- Enable Default Memory Overcommit
- Add GROUPING SETS Support to GROUP BY Clause with Parallel Processing
- Added system.certificates table
- Adds h3Line, h3Distance, and h3HexRing Functions
- New ClickHouse Diagnostics Tool (clickhouse-diagnostics)
- Add Prometheus Output Format Support
- Add MySQLDump Input Format for Table Data Extraction
- Show total_rows and total_bytes for temporary tables in system.tables
- Allow Query-Level Override of Delay and Throw Settings in ClickHouse
Performance Improvement
14- Improve performance of avg and sum without GROUP BY
- Improve Performance of Unary Arithmetic Functions Using Dynamic Dispatch
- Improve ORDER BY and MergeJoin performance with JIT compilation
- Change Structure of system.asynchronous_metric_log to Reduce Space Usage and Remove Redundant Field
- Load Marks for Necessary Columns in Wide Parts
- Improved File Descriptor Cache Performance by Narrowing Mutex Scopes
- Improve performance of File and table functions with globs for large directories
- Apply Parallel Parsing for HiveText to Achieve 2x Speedup
- Default HashJoin Not Thread-Safe; Performance Issues with Large Right Tables
- Allow rewriting countDistinct aggregation to subquery format
- Transform OR LIKE chain to multiMatchAny for improved functionality
- Improve Function Performance through Inlining
- Add Branch to Optimize memcpy in readBig for Improved Performance
- Implement Partial GROUP BY Key for Optimize Aggregation in Order