- Fixed Calcite's JOIN rel to the sql conversion algorithm. It was making wrong decisions about column list replacement into Star.
- Fixed a left-side constant in query condition templates
- ADP only configuration is now valid
- New LL-W functions (UPSERT, DELETE)
- The COLLATE keyword for the WHERE clause in ADG
- MPP-R queries for Logical Materialized Views
- The default application config (with application.yml replacing application-default.yml)
- Timestamp microsecond values for the WHERE clause of LL-R queries
- Improved the performance of query parsing
- Restricted the DDL operations on INFORMATION_SCHEMA
- Removed CORE_TIME_ZONE from the configuration as obsolete
- Refactored JDBC for the date, time and timestamp types algorithms
- Extended the failure logs for MPP-W
- Estimation of an enriched LL-R query by using the ESTIMATE_ONLY hint
- An ability to modify a logical schema without affecting a physical schema by using the LOGICAL_ONLY hint in DDL commands
- Corrected CHAR and UUID logical datatypes names and length in INFORMATION_SCHEMA
- Fixed MPP-R using LIMIT
- Fixed an MPP-R enriched query for ADQM
- Fixed ADQM LL-R JOIN
- Fixed CONFIG_STORAGE_ADD
- Patched JDBC getDate, getTime, getTimestamp to use a specified calendar
- Fixed CONFIG_STORAGE_ADD
- Fixed INFORMATION_SCHEMA.TABLES displaying a logical table datasource after dropping the logical table with the said datasource
- Fixed the recognition of the select category "Undefined type"
- CHECK_SUM and CHECK_DATA can use a normalization parameter for extra-large delta uploads
- Changed the CHECK_SUM and CHECK_DATA summation algorithm
- Included a commons-lang library into JDBC
- Enabled a STACKTRACE logging for blocked threads
- CREATE TABLE in ADQM, ADG, ADP no longer bypasses a check for a sharding key being a subset of a PK
- Changed the CHECK_DATA and the CHECK_SUM parsing to return more specific error messages
- Updated some error messages to be more informational
- New datasource type ADP (PostgreSQL datasource)
- Prerequisites: https://github.com/arenadata/kafka-postgres-connector
- Enabled MPP-R source specification within a respective query
- Added a new column TABLE_DATASOURCE_TYPE to INFORMATION_SCHEMA.TABLES
- Implemented subqueries in a SELECT clause: SELECT * FROM tbl1 WHERE tbl1.id IN (subquery)
- Enabled more than 20 parameters for the IN keyword
- Enriched ADG error message logs (timeouts etc.)
- Changed names of the INFORMATION_SCHEMA sharding keys (TABLE_CONSTRAINTS, KEY_COLUMN_USAGE) to include a datamart name
- Refactored query enrichment for all of the Plugins (ADB, ADG, ADQM) to support the subqueries (see above)
- Fully reworked ADQM enrichment implementation
- Made component names of CHECK_VERSIONS to be consistent with product names
- ROLLBACK DELTA stops running MPP-W operations
- Refactored the stopping mechanizm of MPP-W operations
- SQL+ DML SELECT valid syntax extended for GROUP BY, ORDER BY, LIMIT, OFFSET keywords combination
- Added support for functions within JOIN condition
Fixed start for a configuration without ADG.
-
New logical entity type - automatically synchronized logical materialized view (ADB -> ADG)
- Prerequisites: tarantool-pxf-connector version 1.0 and adg version 0.3.5 or higher
- See the documentation for full information
- Found and eliminated the intermittent ADB MPP-W failure “FDW server already exists”
- JDBC resultset meta-data patched for LINK, UUID logical types
- Patched LL-R query with
count
andlimit
orfetch next N rows only
- Patched ADQM MPP-W for the tables with PK contains Int32 type
- An upgraded Vert.X 4.1, SQL client included
- Enriched logs with the unique operation identifier. The logging template parameter is
%vcl{requestId:-no_id}
- Changed ADB LL-R queries to use Statement instead of PreparedStatement when no parameters supplied
- Added the explicit restriction for duplicate column names for logical entities
- Added a check of a timestamp string format on parsing stage for views and materialized views
- MPP-W failures log info extended
- Updated some error messages to be more informational
- Optimized ADB sql client connection parameters to maximize requests throughput.
- JDBC logging is off by default.
- Query-execution-core new configuration parameters:
executorsCount
: ${ADB_EXECUTORS_COUNT:20}poolSize
: ${ADB_MAX_POOL_SIZE:5}worker-pool
: ${DTM_CORE_WORKER_POOL_SIZE:20}event-loop-pool
: ${DTM_CORE_EVENT_LOOP_POOL_SIZE:20}
- Removed Query-execution-core configuration parameter:
maxSize
: ${ADB_MAX_POOL_SIZE:5}