Skip to content

24.3.11.13
Compare
Choose a tag to compare
@shmel1k shmel1k released this 18 Feb 19:44
· 8548 commits to main since this release
24.3.11.13
68b4343
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Version 24.3.11.13

Release date: December 24, 2024.

Functionality

  • Introduced query tracing, a tool that allows you to view the detailed path of a request through a distributed system.
  • Added support for asynchronous replication, that allows synchronizing data between YDB databases in near real time. It can also be used for data migration between databases with minimal downtime for applications interacting with these databases.
  • Added support for views, which can be enabled by the cluster administrator using the enable_views setting in dynamic configuration.
  • Extended federated query capabilities to support new external data sources: MySQL, Microsoft SQL Server, and Greenplum.
  • Published documentation on deploying YDB with federated query functionality (manual setup).
  • Added a new launch parameter FQ_CONNECTOR_ENDPOINT for YDB Docker containers that specifies an external data source connector address. Added support for TLS encryption for connections to the connector and the ability to expose the connector service port locally on the same host as the dynamic YDB node.
  • Added an auto-partitioning mode for topics, where partitions can dynamically split based on load while preserving message read-order and exactly-once guarantees. The mode can be enabled by the cluster administrator using the settings enable_topic_split_merge and enable_pqconfig_transactions_at_scheme_shard in dynamic configuration.
  • Added support for transactions involving topics and row-based tables, enabling transactional data transfer between tables and topics, or between topics, ensuring no data loss or duplication. Transactions can be enabled by the cluster administrator using the settings enable_topic_service_tx and enable_pqconfig_transactions_at_scheme_shard in dynamic configuration.
  • Implemented Change Data Capture (CDC) for synchronous secondary indexes.
  • Added support for changing record retention periods in CDC topics.
  • Added support for auto-increment columns as part of a table's primary key.
  • Added audit logging for user login events in YDB, session termination events in the user interface, and backup/restore operations.
  • Added a system view with information about sessions installed from the database using a query.
  • Added support literal default values for row-oriented tables. When inserting a new row in YDB Query default values will be assigned to the column if specified.
  • Added support for RETURNING clause in queries.
  • Added start/end times and authors in the metadata for backup/restore operations from S3-compatible storage.
  • Added support for backup/restore of ACL for tables from/to S3-compatible storage.
  • Included paths and decompression methods in query plans for reading from S3.
  • Added new parsing options for timestamp/datetime fields when reading data from S3.
  • Added support for the Decimal type in partitioning keys.
  • Improved diagnostics for storage issues in HealthCheck.
  • (Experimental) Added a cost-based optimizer for complex queries, involving column-oriented tables. The cost-based optimizer considers a large number of alternative execution plans for each query and selects the best one based on the cost estimate for each option. Currently, this optimizer only works with plans that contain JOIN operations.
  • (Experimental) Initial version of the workload manager was implemented. It allows to create resource pools with CPU, memory and active queries count limits. Resource classifiers were implemented to assign queries to specific resource pool.
  • (Experimental) Implemented automatic index selection for queries, which can be enabled via the index_auto_choose_mode setting in table_service_config in dynamic configuration.

YDB UI

  • Added support for creating and viewing information on asynchronous replication instances.
  • Added an indicator for auto-increment columns.
  • Added a tab with information about tablets.
  • Added a tab with details about distributed storage groups.
  • Added a setting to trace all queries and display tracing results.
  • Enhanced the PDisk page with attributes, disk space consumption details, and a button to initiate disk decommissioning.
  • Added information about currently running queries.
  • Added a row limit setting for query editor output and a notification when results exceed the limit.
  • Added a tab to display top CPU-consuming queries over the last hour.
  • Added a control to search the history and saved queries pages.
  • Added the ability to cancel query execution.
  • Added a shortcut to save queries in the editor.
  • Separated donor disks from other disks in the UI.
  • Added support for InterruptInheritance ACL and improved visualization of active ACLs.
  • Added a display of the current UI version.
  • Added a tab with information about the status of settings for enabling experimental functionality.

Performance

  • Accelerated recovery of tables with secondary indexes from backups up to 20% according to our tests.
  • Optimized Interconnect throughput.
  • Improved the performance of CDC topics with thousands of partitions.
  • Enhanced the Hive tablet balancing algorithm.

Bug fixes

  • Fixed an issue that caused databases with a large number of tables or partitions to become non-functional during restoration from a backup. Now, if database size limits are exceeded, the restoration operation will fail, but the database will remain operational.
  • Implemented a mechanism to forcibly trigger background compaction when discrepancies between the data schema and stored data are detected in DataShard. This resolves a rare issue with delays in schema changes.
  • Resolved duplication of authentication tickets, which led to an increased number of requests to authentication providers.
  • Fixed an invariant violation issue during the initial scan of CDC, leading to an abnormal termination of the ydbd server process.
  • Prohibited schema changes for backup tables.
  • Fixed an issue with an initial scan freezing during CDC when the table is frequently updated.
  • Excluded deleted indexes from the count against the maximum index limit.
  • Fixed a bug in the display of the scheduled execution time for a set of transactions (planned step).
  • Fixed a problem with interruptions in blue–green deployment in large clusters caused by frequent updates to the node list.
  • Resolved a rare issue that caused transaction order violations.
  • Fixed an issue in the EvWrite API that resulted in incorrect memory deallocation.
  • Resolved a problem with volatile transactions hanging after a restart.
  • Fixed a bug in the CDC, which in some cases leads to increased CPU consumption, up to a core per CDC partition.
  • Eliminated read delays occurring during and after the splitting of certain partitions.
  • Fixed issues when reading data from S3.
  • Corrected the calculation of the AWS signature for S3 requests.
  • Resolved false positives in the HealthCheck system during database backups involving a large number of shards.
Assets 2
Loading