The Valkey Bloom module is now available on Aiven for Valkey services running Valkey 9 and later. It adds a probabilistic Bloom filter data type that lets you test set membership using very little memory. A Bloom filter can tell you that an element is possibly in a set or definitely not in a set, which makes it well suited to high-traffic use cases such as preventing cache penetration, deduplicating streams, and filtering out known items.

Key features

• Space-efficient membership tests: Check whether an element is likely present or definitely absent using a fraction of the memory needed to store the elements themselves.
• Auto-scaling filters: Filters expand automatically as you add elements, preserving the configured false positive rate.
• Configurable accuracy: Set a custom capacity and false positive rate per filter.
• Persistence: Bloom filters are saved and restored together with the rest of your Valkey data.

Configuration

Valkey Bloom is enabled by default on all Aiven for Valkey services running Valkey 9 and later. No configuration is required, and the module cannot be disabled.

The following BF.* commands are available immediately: BF.ADD, BF.EXISTS, BF.MADD, BF.MEXISTS, BF.RESERVE, BF.INSERT, BF.CARD, and BF.INFO.

Getting started

• For existing services: Apply the available maintenance update to activate Valkey Bloom, then start creating filters with BF.RESERVE and checking membership with BF.ADD and BF.EXISTS.
• For new services: Valkey Bloom is enabled automatically when you create a new Aiven for Valkey service running Valkey 9 or later.

Learn more

• Aiven for Valkey modules reference — Aiven-specific configuration and supported commands
• Valkey Bloom documentation — Open-source Valkey Bloom docs, including commands, scaling behavior, configuration options, and memory management

The Valkey Search module is now available for all Aiven for Valkey services. It extends Valkey with a high-performance search engine that supports vector search, full-text search, and hybrid queries over data stored in Valkey Hash and Valkey JSON.

Key features

• Vector search: Run Approximate Nearest Neighbor (HNSW) or exact K-Nearest Neighbor (KNN) searches with low latency.
• Hybrid search: Combine vector queries with numeric, tag, and full-text filters in a single query.
• Flexible indexing: Create indexes over Valkey Hash and Valkey JSON data types.
• Cluster support: Fully supported on Aiven for Valkey cluster plans.

Configuration

Valkey Search is enabled by default on all Aiven for Valkey services. No configuration is required, and the module cannot be disabled.

The following FT.* commands are available immediately: FT.CREATE, FT.SEARCH, FT.AGGREGATE, FT.DROPINDEX, FT.INFO, and FT._LIST.

Getting started

• For existing services: Apply the available maintenance update to activate Valkey Search, then start creating indexes with FT.CREATE and querying them with FT.SEARCH.
• For new services: Valkey Search is enabled automatically when you create a new Aiven for Valkey service.

Learn more

• Aiven for Valkey modules reference — Aiven-specific configuration and supported commands
• Valkey Search documentation — Open-source Valkey Search docs, including query syntax, expressions, data formats, and monitoring

The organization:billing:write permission lets users manage billing information and payment methods. For users who only need to view payment details and invoices, the organization:billing:read permission grants read-only access.

More information on permissions is available in the documentation.

Arrow Flight connection details are not yet available in the Aiven Console. To get the connection details, use the Aiven API, Aiven CLI, or Aiven MCP server.

For more information, see Supported interfaces and drivers.

For more details on the supported features and limitations, see the Aiven for OpenSearch Developer tier page or the docs.

Supported functionality:

• Write Kafka records to Salesforce objects
• Insert records using Salesforce Bulk API 2.0
• Support schema-based and schemaless record values

Key considerations

• Insert operations only. Updates, upserts, and deletes are not supported.
• At-least-once delivery. Duplicate records can occur.
• Kafka record fields must match Salesforce field API names.

For configuration and usage details, see the Salesforce sink connector documentation.

Schema references let you register schemas that depend on other schemas already stored in Schema Registry. This helps you reuse shared definitions across subjects instead of copying the same schema definitions into each schema.

Karapace also supports schema references for Protobuf schemas. Schema references are not supported for JSON Schema.

For more information, see Schema references in Karapace.

See PostgreSQL release notes for more details about the PostgreSQL minor releases.

• Added retry-and-reconcile handling for concurrent schema update failures in GCS to BigQuery load jobs.
• Added write-attempt metadata with putAttemptId, which helps downstream systems distinguish rows written by different Kafka Connect put() attempts.
• Updated dependency versions, including Jackson, Jetty, and Apache Kafka clients.

For more information, refer to the GitHub release notes for the BigQuery connector for Apache Kafka.

Recommended actions

Manually upgrade your services before the EOL date to ensure a controlled transition:

1. Test via forking: Use the Aiven fork feature to clone your 14 service into a target-version instance. This allows you to verify application compatibility without affecting production data.
2. Check for breaking changes: Review breaking changes between the versions, if any.
3. Perform manual upgrade: Follow the upgrade guide to trigger the version change at your convenience.

Customer managed keys are now supported across all service resources through the new cmk_id field. This makes it easier to apply bring-your-own-key encryption consistently across services.

This release also introduces important compatibility updates:

• BYOC on AWS plans are now more accurate: changes that require new infrastructure will recreate the resource instead of attempting an in-place update.
• Expanded service configuration support:
  • ClickHouse, MySQL, and DataHub integrations have new configuration support.
  • Kafka and Kafka Connect now have IPv6 connector resolution, Vault certificates, and safer handling of environment secrets.

Cassandra and Redis resources and data sources have also been removed in this release.

For full details, see the complete changelog.

The AMQP source connector reads messages from AMQP-compatible queues, such as RabbitMQ, and writes them to Apache Kafka® topics. Each source message is serialized as JSON and written as a Kafka record with a generated ULID key.

Key considerations:

• Supports AMQP queues only.
• key.converter must use StringConverter.
• value.converter must use StringConverter or JsonConverter.
• For single-queue deployments, tasks.max=1 is recommended unless multiple consumers are intentionally configured.
• Messages available while the connector is offline can be missed.
  
  

For configuration and usage details, see the AMQP source connector documentation.

What this means for you:

• New services: When creating a new Aiven for Valkey™ service, the platform will now automatically select version 9 as the default.
• Existing services: Your currently running services on version 8.1 remain unaffected and will continue to operate normally.

See our documentation for the Aiven for Valkey service version lifecycle.

• ap-melbourne-1: Australia, Melbourne
• eu-milan-1: Italy, Milan
• eu-turin-1: Italy, Turin
• me-jeddah-1: Saudi Arabia, Jeddah
• me-riyadh-1: Saudi Arabia, Riyadh

The OCI regions support is a limited availability feature. For more information or access, contact your account team.

See all the supported OCI regions or all cloud providers and regions supported on the Aiven Platform.

• GET /v1/project: Use GET /v1/organization/{organization_id}/projects instead.
• DELETE /v1/project/{project}: Use DELETE /v1/organization/{organization_id}/project/{project_id} instead.
• GET /v1/project/{project}: Use GET /v1/organization/{organization_id}/project/{project_id} instead.
• PUT /v1/project/{project}: Use PATCH /v1/organization/{organization_id}/project/{project_id} instead. The replacement is a PATCH endpoint, which changes this from full replace to partial update.
• GET /v1/account/{account_id}/projects: Use GET /v1/organization/{organization_id}/projects instead.

If you previously relied on a single global listing call using GET /v1/project, you now need to iterate organizations and call the organization-scoped listing endpoint per organization.

For more information, see the Aiven API documentation.

LTR is available for Aiven for OpenSearch® 2.19+ and 3.3+.

For details on all TimescaleDB releases, see the release notes.

What this means for you:

• New services: All new Aiven for Dragonfly services provisioned moving forward will automatically run on version 1.34.2.
• Existing services: Existing instances will be scheduled for a rolling upgrade as part of our standard maintenance cycle.

For a full list of changes included in Dragonfly releases, refer to the official Dragonfly release notes.

• Managed Kafka service with Schema Registry and REST Proxy
• Higher throughput than the Free tier
• Configurable message retention: 1, 2, or 3 days
• Optional Kafka Connect add-on
• Integrations and monitoring

Pricing

• Starts at $35 USD/month for 1 day retention
• Kafka Connect service starts at $20 USD/month
• Extending retention beyond the default increases the monthly cost

Limitations

• Designed for development, testing, and smaller workloads
• Resource and usage limits apply
• Some advanced features are available only on the Professional tier

Upgrade to a Professional tier plan to remove limits and enable production features.

For more information, see the Aiven for Apache Kafka Developer tier documentation.

The 8.4 LTS release comes with the following key changes:

• Safer authentication defaults: mysql_native_password is disabled by default, which reduces reliance on legacy authentication.
• Stricter schema validation during upgrades: deprecated patterns such as AUTO_INCREMENT on FLOAT and DOUBLE are no longer supported, helping catch risky definitions early.
• Better optimizer maintenance options: support for automatic histogram updates helps keep query planning statistics fresh.
• Replication and operations improvements: updates such as GTID tagging and retry behavior changes improve tracking and operational tuning for replicated setups.

For complete details, see the official MySQL 8.4 release notes.

• POST /v1/project/{project}/invite: Use POST /v1/organization/{organization_id}/invitation instead
• POST /v1/project/{project}/invite/{invite_verification_code}: Use POST /v1/organization/{organization_id}/invitation/weblink/{invite_verification_code} instead
• GET /v1/project/{project}/users: Use GET /v1/organization/{organization_id}/permissions/project/{project_id} instead
• PUT /v1/project/{project}/user/{user_email}: Use PATCH /v1/organization/{organization_id}/permissions/project/{project_id} instead
• DELETE /v1/project/{project}/user/{user_email}: Use PATCH /v1/organization/{organization_id}/permissions/project/{project_id} instead

For more information, see the Aiven API documentation.

For complete migration details and requirements, see the documentation for KRaft in Aiven for Apache Kafka.

For details on all TimescaleDB releases, see the release notes.

For details, see:

• Apache Kafka® 4.1.0 release notes
• Apache Kafka® 4.1.1 release notes
  

This setting controls how ClickHouse's SLRU cache splits between protected and probationary segments for HNSW vector similarity indexes. The Aiven default is `0.1`. Changing this setting doesn't require a service restart.

• Retrieve data in batches using incremental sync with LastModifiedDate
• Run multiple SOQL queries
• Route records to topics using topic.prefix (per object) or topic (single topic)

Limitations

• Single task only (tasks.max=1)
• Polling-based ingestion; no real-time streaming
• At-least-once delivery; duplicate records can occur

For configuration and usage details, see the Salesforce source connector documentation.

• azure-mexico-central — North America, Mexico: Mexico Central
• azure-indonesia-central — Asia, Indonesia: Central Indonesia

These new regions enable you to create Aiven-managed services closer to your users in Mexico and Indonesia for improved latency and compliance.

For the complete list of available cloud regions, see the list of clouds documentation.

It's ideal for filtering data streams in real time, building anomaly detection, creating admin dashboards, running LLMs and AI agents securely, or shipping other internal tools in a secure and scalable way.

Aiven Apps is currently in limited availability. To request access, contact Aiven. For more information, see the Aiven Apps documentation.

This feature is in limited availability.

• Migration window: Monday to Friday, 00:00 to 14:00 UTC
• Support added for business-16 and business-32 service plans

For more information, see KRaft mode in Aiven for Apache Kafka®.

• Beauharnois (BHS1)
• Beauharnois (BHS3)
• Beauharnois (BHS5)
• Frankfurt (DE1)
• Gravelines (GRA1)
• Gravelines (GRA3)
• Gravelines (GRA5)
• Gravelines (GRA7)
• Gravelines (GRA9)
• Gravelines (GRA11)
• Milan (MIL)
• Paris (PAR)
• Strasbourg (SBG5)
• Strasbourg (SBG7)
• Singapore (SGP1)
• London (UK1)
• Warsaw (WAW1)

Check all cloud regions available on Aiven in the documentation.

What's new:

• Save queries: You can save frequently used SQL queries for quick access and reuse. Saved queries are user-specific and private within an Aiven for PostgreSQL® service.
• Manage saved queries: You can rename or delete saved queries directly from the editor interface.

Benefits:

• Reduces query rewriting time during repeated analysis sessions
• Improves workflow efficiency with quick access to historical queries and saved analysis

Availability: Available in PG Studio for Aiven for PostgreSQL services
Learn more: Manage and persist queries in PG Studio

Currently supported versions include:

• ClickHouse 25.3 (default)
• ClickHouse 25.8

Downgrading to an earlier ClickHouse version is not supported.

For more information, see Manage versions in Aiven for ClickHouse®.

• Vector search GA with binary quantization for lower memory usage and faster indexing
• Lightweight projections act as secondary indexes without duplicating data
• Lightweight updates (beta) — UPDATE ... SET ... WHERE on MergeTree tables
• Correlated subqueries and lag/lead window functions for broader SQL compatibility
• Iceberg and Delta Lake improvements including write support, catalog integration, and position/equality deletes
• Parquet reader v3 with filter pushdown for faster queries on external files
• Lazy materialization and query condition cache for improved query performance

More engines supported

• MergeTree family
  • CoalescingMergeTree
  • ReplicatedCoalescingMergeTree
• Integration
  • GCS
• Data lake
  • DeltaLakeAzure
  • DeltaLakeS3

Table engines supported in ClickHouse 25.3 remain available in ClickHouse 25.8. For the complete list of upstream changes, see ClickHouse 2025 Changelog.

Key features

• Fully managed Aiven for OpenSearch® service at no cost forever
• 4 GB RAM, 20 GB storage
• Sample data in OpenSearch Dashboards for testing indexing and search
• Basic monitoring for metrics and logs
• Standard Aiven for OpenSearch clients supported

Use cases

• AI-powered search
• Retrieval-augmented generation (RAG)
• Log analytics

Limitations

• One free Aiven for OpenSearch service per organization
• Fixed limits on storage, performance, and concurrent connections
• Other limitations on service functionality and advanced features

To learn more, see the Aiven for OpenSearch free tier documentation.
You can upgrade to a paid plan at any time to remove limits and enable additional features.

Limitation: Classic Kafka services cannot be upgraded to Inkless at this time.

For more information, see the Inkless documentation.

Configuration is available through the Aiven API, Aiven CLI, and Aiven Provider for Terraform. The feature is not available in the Aiven Console.

For more information, see Configure secret providers.

Key features include:

• Emissions estimates for services running on AWS, Microsoft Azure, and Google Cloud
• A carbon emissions explorer with filtering by time range, billing groups, and projects
• GDPR-compliant data processing within Aiven's environment

Learn more:

• All supported UpCloud regions
• All supported cloud providers and regions

For more details and upgrade instructions, see our documentation.

Key highlights

• Hash field expiration: You can now set TTLs (time-to-live) on individual fields within a single Hash key using commands like HEXPIRE and HTTL. This reduces memory overhead by allowing specific pieces of data to expire without needing to delete the entire key or split data across multiple keys.
• Major throughput boosts:
  • Zero-copy & prefetching: Up to 40% higher throughput for pipelined requests and 20% for large responses.
  • Multipath TCP: Reduces network latency by up to 25%.

Get started

You can now select version 9.0 when creating a new service or upgrade your existing service directly in the Aiven Console.

Configured switchover windows apply to maintenance updates started automatically during the maintenance window and to manually started maintenance updates. This gives you more predictable maintenance timing and helps reduce operational impact.

For more information, see Controlled maintenance updates in Aiven for PostgreSQL®.

Key features

This early availability release features the following:

• Natural language to SQL: Generate precise SQL by describing your goal in plain English. The AI uses your unique schema map to ensure queries are context-aware and accurate.
• Interactive schema mapping: Visualize your database structure through a dynamic map of tables, columns, and relationships, making it easy to navigate complex data models.
• Query insights & explanations: Get instant breakdowns of existing queries, complex joins, or unfamiliar database objects to speed up troubleshooting and onboarding.
• Zero configuration: Available instantly for all Aiven for PostgreSQL® services—optimized for fast debugging, data exploration, and ad-hoc analysis.

Security and stability guardrails

To protect your production environments, PG Studio includes built-in enterprise-grade safeguards:

• Privacy-first AI: Your data remains yours. We use only schema metadata for context; your actual database content is never used to train AI models.
• Automatic safety limits: Includes single-statement validation and a 30-second query timeout to prevent runaway processes and maintain service stability.

To get started, explore our public documentation and check out the Introducing Aiven’s PG Studio blog post.

• Upgrading a service from Kafka 3.8 to Kafka 3.9 automatically migrates the service from ZooKeeper to KRaft.
• The upgrade involves rolling restarts, which may cause temporary client reconnections.
• ZooKeeper-related controller metrics are no longer available after migration.
• Migration rollout includes temporary limitations such as supported service plans, migration windows, and limits on concurrent migrations.
• Migrating to KRaft is required before upgrading to Apache Kafka 4.0 or later.

For details, see KRaft in Aiven for Apache Kafka®.

For details on all TimescaleDB releases, see the release notes.

For details on all TimescaleDB releases, see the release notes.

The following parameters can now be adjusted both up and down:

• max_connections
• max_locks_per_transaction
• max_prepared_transactions
• max_wal_senders
• max_worker_processes

Decreasing these parameters typically requires a service restart to apply the changes. Ensure you have a maintenance window or high-availability (HA) enabled before proceeding.

Learn more

Check out the full list of Aiven for PostgreSQL advanced parameters in our documentation for details on default values and limits.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

• Improves cluster scaling, rebalancing, and recovery compared to classic Kafka with tiered storage.
• Supports both classic and diskless topics within the same Kafka service.
• Stores retained data in object storage, reducing the need to pre-provision local disk for long-term retention.
• Uses Apache Kafka 4.0.

Limitations

• Available only for new services; Classic Kafka services cannot be upgraded to Inkless at this time.

For more information, see the Inkless overview documentation.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

Other key improvements in this release include:

• Enhanced Valkey user management: ACL fields (valkey_acl_categories, valkey_acl_commands, valkey_acl_keys, valkey_acl_channels) can now be updated in-place without forcing resource recreation
• Improved Kafka topic handling: Better support for infinite retention with retention_bytes = -1
• OpenSearch 3.3 support: The latest OpenSearch version is now available

This release also includes various bug fixes for MySQL user password updates, service cloud name handling, and billing group configurations. For complete details on all fixes and updates, view the full Aiven Provider for Terraform changelog.

Impact:

If a connector uses credential_source.url, configure gcp_auth_allowed_urls at the Kafka Connect service level. Connectors using file- or command-based credential sources fail validation.

More information:

• Create a Google Cloud Storage sink connector for Apache Kafka®
• Create a sink connector from Apache Kafka® to Google BigQuery

Key changes include:

• Improved backup and startup times: Faster backups lead to shorter maintenance windows and significantly reduced service startup times.
• Enhanced HA handling: The architecture provides greater resilience against network partitions and outages, and a new API allows you to trigger manual switchovers.
• Rapid multi-node deployment: Optimized processes allow for faster startup times for multi-node service plans.
• Streamlined pg_upgrade experience: Standby nodes now upgrade in-place, eliminating the wait times for multi-node services to become fully available after an upgrade.
• Maintenance efficiency: Parallel node provisioning reduces the total time required for maintenance and synchronization for multi-node service plans.

For details on all TimescaleDB releases, see the release notes.

For details on all TimescaleDB releases, see the release notes.

For details on all TimescaleDB releases, see the release notes.

For Aiven for Apache Kafka® MirrorMaker 2, the follower_fetching_enabled field was added to the aiven_mirrormaker_replication_flow resource letting you enable rack awareness and follower fetching.

For the complete list of changes and technical details, view the full Aiven Provider for Terraform changelog.

When all standby nodes are unavailable, read traffic routed to the HA replica DNS endpoint is automatically redirected to the healthy primary node, ensuring uninterrupted access and minimizing downtime.

Key highlights:

• Seamless failover for read-only queries when standbys are unavailable
• No application-side routing changes required
• Single, stable connection endpoint for read traffic
• Easy enable/disable via Console, CLI, API, Terraform, or Kubernetes

To learn more, see the feature documentation.

Key timeline

• End-of-Life (EOL) date: July 26, 2026
• Automatic upgrades: Starting July 26, 2026, any services remaining on version 2.17 will be automatically upgraded to version 2.19 during their next scheduled maintenance window.

Why upgrade to 2.19?

Upgrading to version 2.19 is a proactive step for your infrastructure:

• Path to major version 3: Version 2.19 serves as the essential "jump version" required for the future upgrade to OpenSearch major version 3, which will bring significant performance improvements.
• Long-term stability: Version 2.19 is the current stable standard on the Aiven platform, offering the latest bug fixes and features over 2.17.

Recommended actions

We recommend manually upgrading your services before the July deadline to ensure a controlled transition:

1. Test via forking: Use the Aiven fork feature to clone your 2.17 service into a 2.19 instance. This allows you to verify application compatibility without affecting production data.
2. Review documentation: Check the OpenSearch breaking changes guide for any minor breaking changes between 2.17 and 2.19.
3. Perform manual upgrade: Follow our upgrade guide to trigger the version change at your convenience.

Key timeline

• End-of-life (EOL) date: July 26, 2026
• Automatic upgrades: Starting July 26, 2026, any services remaining on version 1.x will be automatically upgraded to version 2.19 during their next scheduled maintenance window.

Why upgrade?

Upgrading to OpenSearch 2.x provides significant technical advantages:

• Enhanced performance: Version 2.x is approximately 6x faster than version 1.3.
• Reduced storage footprint: Zstandard compression reduces on-disk storage requirements by 15% to 30%.

Recommended Actions

To ensure a smooth transition and verify application compatibility, we recommend taking the following steps before the July deadline:

1. Review breaking changes: Consult the OpenSearch 2.0 breaking changes guide to identify potential impacts.
2. Test via forking: We suggest forking your service at its current version, then upgrading the fork to version 2.19 to test for incompatibilities in a safe environment.
3. Perform manual upgrade: Upgrade your services at your convenience before the EOL date.

Highlights

• OpenSearch Dashboards - redesigned experience
  • Unified observability experience (preview): A redesigned Discover interface in OpenSearch Dashboards brings log analytics and distributed tracing together, offering an all-in-one view for troubleshooting and monitoring. To ensure service stability, the new Discover experience is disabled by default while in beta for version 3.3. We plan to enable this by default once it reaches general availability (GA) in a future release. To enable this feature before its GA, contact your account manager.
  • Smarter visualizations: New features enable smarter, automated visualizations and more chart types, making data analysis faster and insights easier to obtain.
  • Discover traces: A dedicated, central interface for querying and exploring traces across your distributed systems.
• Enhanced performance and core search
  • Lucene 10 upgrade (since v3.0): This core upgrade modernizes the search infrastructure, leading to improved indexing and search performance, and better efficiency.
  • Better hybrid search: Enhancements like Z-score normalization for result scoring provide more accurate and relevant results when combining different search methods.
  • Star-tree indexing improvements: Speeds up slow-running aggregations, particularly multi-terms aggregations, for large datasets.
• Powerful analytics with piped processing language (PPL)
  • Apache Calcite integration (since v3.0): PPL now uses the Apache Calcite query engine by default, introducing better portability, optimization capabilities, and an extensive library of new PPL functions for data manipulation and analysis.
  • New PPL commands: Introduction of commands like timechart for time-series aggregation and enhanced functions like coalesce and mvjoin for complex data transformations.
• Vector search enhancements: Continued improvements to vector search, including the new Seismic sparse retrieval algorithm for faster neural sparse search, supporting more sophisticated generative AI and semantic search applications.

Supported upgrade paths

• 1.3 → 2.19 → 3.3
• 2.17 → 2.19 → 3.3

To explore all changes and make your OpenSearch experience even better, review the following:

• Full release notes
• Breaking changes

A major change in this version is the complete removal of the Redis kind and controller from the operator, as Redis has reached end of life. Use Valkey instead for all caching services.

Key enhancements in this release include:

• Enhanced backup capabilities: Both Kafka and PostgreSQL now support configurable automatic backups with new fields for backup intervals and retention periods
• Improved authentication: Kafka services now support Let's Encrypt certificate authorities for SASL authentication and SASL OAUTHBEARER with configurable allowed URLs
• MySQL migration improvements: Added experimental database dump tool configuration and replication options for enhanced migration workflows
• PostgreSQL high availability: New dedicated read-only DNS option that automatically falls back to primary when standby nodes are unavailable
• Performance optimizations: Valkey I/O threads maximum increased from 32 to 256, and various field limits expanded

For complete details on all changes and deprecations, check out the full Aiven Operator changelog.

Kafka Connect and MirrorMaker 2 automatically use the availability zone where each node runs to prefer reading from in-sync follower replicas in the same zone. This reduces cross-availability zone network traffic and associated costs.

For Kafka Connect, this applies to sink connectors only. For MirrorMaker 2, follower fetching is enabled by default for new replication flows and can be disabled per flow.

Learn more:

• Follower fetching in Aiven for Apache Kafka®
• Configure rack awareness in Aiven for Apache Kafka® MirrorMaker 2

Supported functionality:

• Change data capture (CDC) for Oracle tables using LogMiner.
• Row-level inserts, updates, and deletes.
• Integration with Schema Registry and Avro serialization.

Limitations:

• Large initial snapshots are not supported. The connector streams ongoing changes only.
• Exactly-once delivery is not supported.

To learn how to configure and use the connector, see the Oracle Debezium source connector documentation.

Other notable improvements in this release:

• Enhanced Kafka topic handling with retry logic for 404 errors from the KafkaTopicListV2 endpoint
• Added mysql_user_config.migration.reestablish_replication field for Aiven for MySQL services to skip dump-restore and start replication directly
• Improved error handling for resources that weren't properly removing from state when deleted externally

Additionally, the termination_protection field for aiven_mysql_database has been deprecated in favor of using Terraform's built-in prevent_destroy lifecycle rule.

For more details on all changes and fixes, view the complete Aiven Provider for Terraform changelog.

The Valkey™ JSON module is now available for all Aiven for Valkey™ services. This module extends Valkey features with advanced JSON data handling capabilities.

Key features

• Native JSON storage: Store JSON documents as values directly in Valkey
• JSONPath queries: Query JSON documents using standard JSONPath syntax
• Atomic updates: Make atomic updates to JSON elements for consistency

Configuration

Enable the module on your existing services by applying a maintenance update.

Otherwise, no action is required from you:

• Enabled by default on all new services.
• No setup required; start using JSON commands immediately.
• Always available; cannot be turned off.

Getting started

• For existing services: Apply a maintenance update to enable and start using the module.
• For new services: Start using Valkey JSON commands right away.

Learn more

• For complete information on Valkey JSON commands and usage examples, visit the Valkey™ JSON documentation.
• For information on Aiven for Valkey, visit:
  • Aiven for Valkey™ modules reference
  • Aiven for Valkey™ documentation

Highlights

• S3Queue table engine
• AzureQueue table engine
• New query analyzer
• Improved JOIN performance
• Data types: JSON, Variant, Dynamic
• Refreshable materialized view
• New generateSerialID function

For all the introduced changes, see relevant sections of the ClickHouse changelog (24.9, 24.10, 24.11, 24.12, 25.1, 25.2, 25.3).

For details, see Apache Kafka® 4.1.0 and 4.1.1 release notes:
- Kafka 4.1.0 release notes
- Kafka 4.1.1 release notes

By default, the value from sasl_oauthbearer_jwks_endpoint_url is added automatically to the allowed list. You can explicitly allow additional URLs by setting sasl_oauthbearer_allowed_urls.

URLs that are not included in either sasl_oauthbearer_allowed_urls or sasl_oauthbearer_jwks_endpoint_url are now blocked for SASL OAUTHBEARER authentication.

Although Apache Kafka Connect and Apache Kafka MirrorMaker 2 do not officially support SASL OAUTHBEARER, this change prevents unsupported configurations from bypassing security controls.

For more information, see:

• sasl_oauthbearer_allowed_urls configuration reference: Advanced parameters for Aiven for Apache Kafka®
• CVE-2025-27817: NVD - CVE-2025-27817
  

Additionally, a new aiven_organization_user_group_member_list data source has been added. It lists all members of a specific user group, making group and permission management easier.

This release also includes regular updates to service configurations, integrations, and endpoints. For more details on fixes and updates, view the complete Aiven Provider for Terraform changelog.

A free tier is now available for Aiven for Apache Kafka. No credit card is required.
Use it for learning, prototyping, small-scale development, and workloads that fit within the free tier limits.

Key features

• Managed Kafka cluster with Karapace Schema Registry and REST Proxy
• Throughput limits of up to 125 KiB/s for produce traffic and 125 KiB/s for consume traffic
• Up to 5 topics, each with two partitions
• Sample data generation and basic monitoring
• One free Kafka service per organization

Limitations

• Free tier services are automatically powered off during extended periods of inactivity and can be reactivated at any time.
• Kafka Connect, MirrorMaker 2, service integrations, and tiered storage are not available.
• Data retention is fixed, and no SLA coverage is included.

You can upgrade to a paid plan at any time to remove limits and enable additional features. To learn more, see the Kafka free tier documentation.

Aiven for Valkey clustering automatically partitions your data across multiple nodes (shards), enabling you to handle larger datasets and higher traffic loads than a single-node deployment.

Key benefits:

• High availability: Automatic failover, minimal downtime, and replica promotion to maintain service continuity during unexpected failures or maintenance.
• Scalability & performance: Distribute read and write operations across multiple shards for significantly higher throughput, lower latency, and better load distribution.
• Compatibility: Fully compatible with existing Valkey cluster-aware client libraries for seamless integration with current applications.
• Flexible sizing: Supports various instance sizes, including smaller 4 GB RAM instances, allowing for better cost optimization.

Learn more in the Aiven for Valkey™ clustering documentation.

Key highlights:

• Seamless failover for read-only queries when standbys are unavailable
• No application-side routing changes required
• Single, stable connection endpoint for read traffic
• Easy enable/disable via Console, CLI, API, Terraform, or Kubernetes

This is a limited availability feature. To learn more, see the feature documentation.

When you enable this option, the connection details display the appropriate SASL port for your access route. Public CA support is available for all connection types, including PrivateLink.

To enable Public CA, use the Aiven Console, Aiven CLI, API, or Terraform. For more information, see Enable public CA for SASL authentication.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

Improved

• Protobuf schemas now use absolute runtime paths when compiled, ensuring that generated *_pb2.py modules load correctly across different runtime environments.

Fixed

• Fixed protobuf decoding failures that caused messages to appear empty or not load in the Kafka Topic Catalog and REST API.
• Fixed an error where protobuf keys that could not be decoded resulted in an Unprocessable Entity response. Karapace now falls back to binary decoding for the key.
• Fixed errors that caused protobuf-formatted messages to fail when fetched through the API or the Aiven Console.

For more information, see the Karapace 5.0.3 release notes.

Performance improvements

• New hashtable implementation: Reduces the average memory needed for key-value pairs and the number of random memory accesses. It also improves throughput by approximately 10% on pipeline workloads when not using I/O threads.
• Iterator prefetching: Up to 3.5x faster when iterating over the key set by using prefetching techniques.
• Various command optimizations: Commands such as ZRANK, ZADD, PFMERGE, PFCOUNT, and BITCOUNT were optimized to execute faster.

Observability improvements

• Extending the slowlog to commandlog: The new COMMANDLOG feature gives you better visibility into end-to-end latency.
• Improved latency insights: The new LATENCY LATEST command reports on the latest collected latency events.

Other improvements

• Conditional updates: The introduction of conditional updates using the SET IFEQ semantics.

Upgrade to version 8.1 at your earliest convenience to ensure you use the most robust and secure version available.

See the Valkey 8.1 release notes for more details.

Key changes include the following:

• Added support for the BigQuery metastore catalog to integrate with Google BigQuery.
• Added Google Cloud authentication support, including Application Default Credentials and service account key files.
• Added support for multiple Google Cloud Storage credential prefixes, allowing different credentials per bucket or prefix.
• Added a new KeyManagementClient for Google Cloud KMS to manage encryption key wrapping and unwrapping.

For more details, see the Apache Iceberg 1.10.0 release notes.

To learn how to use the connector, see the Iceberg sink connector documentation.

There are also 3 new data sources that help you get details without using another tool like the Aiven Console. The aiven_service_plan and aiven_service_plan_list data sources give you detailed service plan information. You can use these to get information on the available service plans.

Additionally, the aiven_organization_user_group_list data source lists all user groups within an organization. This makes it easier to manage user groups and assign them permissions.

This release also includes regular updates to service configurations, integrations, and endpoints. For more details on fixes and updates, view the complete Aiven Provider for Terraform changelog.

More information on the Developer plan is available in the Aiven docs.

Learn more about how Aiven for OpenSearch BTAR works and see how to enable this limited availability feature.

• Added an option to skip fields not defined in the BigQuery schema.
• Added support for partition decorator syntax in the Storage Write API.
• Added support for updating table schemas in GCS-to-BigQuery load jobs.
• Enabled connection pooling for the Storage Write API default stream to improve write performance.
• Added trace ID propagation to help BigQuery trace connector traffic.

For configuration instructions, see Create a sink connector from Apache Kafka® to Google BigQuery

For all releases, see the GitHub repository.

How this change applies

• New services (affected): New Aiven for PostgreSQL services in organizations without PGBouncer pools will use scram-sha-256 by default.
• Existing services (unaffected): All existing services are not impacted by this change.
• PGBouncer pool organizations (unaffected): New services created in organizations with PGBouncer pools will continue to default to md5.

Client compatibility: Most modern PostgreSQL client libraries are compatible with scram-sha-256. For authentication issues, consult your driver's documentation or the PostgreSQL list of drivers.

When to migrate manually

If your organization has PGBouncer pools tied to specific database users or your services have additional database users created, migrate to scram-sha-256 manually before the next major PostgreSQL version upgrade, when md5 encryption will be deprecated.

View migration guide and compatibility instructions

How to troubleshoot authentication issues

If you experience connection problems with scram-sha-256, you have two immediate options:

• Revert default: Change pg.password_encryption back to md5 in your service's user_config.
• Role-specific MD5: For specific roles, create them with md5 secrets directly using CREATE ROLE ... WITH PASSWORD 'md5' ... (PostgreSQL documentation on CREATE ROLE).

This particular vulnerability could allow an authenticated user to obtain remote code execution using a specially crafted Lua script.

This affects all Redis and Valkey versions that have Lua scripting support.

This issue is resolved in:

• Redis: 8.2.2+, 8.04+, 7.4.6+, 7.2.11+, 6.2.20+
• Valkey: 8.1.4+, 8.0.6+, 7.2.11+

On October 8th Valkey 8.0.6 was made available as mandatory maintenance on the Aiven platform. It is recommended that all Valkey customers run this maintenance as soon as possible.

For those customers running Redis (Aiven for Caching) on the Aiven platform, a patch to 7.2.11 will be released in the coming days. Note that Aiven for Caching is end-of-life on the Aiven platform and all running instances will be decommissioned starting on October 15th, 2025.

• Data durability and replication are handled by the object storage layer.
• Flexibility of classic and diskless topics both coexisting on the same cluster.
• Runs a PostgreSQL-based coordinator, managed as a service integration.

Learn more in the Diskless topics overview.

Available for Aiven for OpenSearch 2.4+ and OpenSearch Dashboards 2.19+. Learn more

For more information, see the Aiven Kafka Connect S3 source connector v3.4.1 release notes.

View all Aiven for PostgreSQL extensions

View instructions for assigning billing groups in the Aiven docs.

Here are some of the key highlights you can start using today:

• Asynchronous I/O subsystem: Significantly boosts performance for sequential scans, bitmap heap scans, and vacuums by queuing multiple read requests.
• Faster joins and GROUP BY: Experience improved performance and lower memory usage for these common operations.
• Parallel GIN indexes: Create GIN indexes in parallel to dramatically reduce maintenance time.
• NOT ENFORCED constraints: A new option that allows you to add CHECK and foreign key constraints without them being enforced, useful for data migration and schema evolution.
• Skip scans for B-tree indexes: This new capability improves performance for specific queries by allowing the index to "skip" over irrelevant data.
• Improved locking performance: Enjoy better performance in high-contention databases as queries that access many relations now have improved locking.

Learn more about this release and what it means for you:

• Full PostgreSQL 18 release notes
• Aiven for PostgreSQL service version lifecycle

Key changes

• New consumer group protocol
  • Kafka 4.0 introduces an improved consumer group coordination protocol.
  • Older clients that do not support this protocol cannot connect to Kafka 4.0 clusters.
  • Upgrade client libraries before connecting.
• Deprecated APIs and protocols removed
  • Kafka 4.0 removes API versions and protocol features that were previously deprecated.
  • Ensure all clients and tooling are compatible before upgrading.
  • This includes Kafka Connect, MirrorMaker 2, and custom client applications.
  • For details, see the Kafka 4.0 release notes.
• `message.format.version` setting removed
  • Kafka 4.0 no longer uses the `message.format.version` topic-level setting.
  • Any configured value is ignored and removed during topic creation or update.
• Timestamp difference setting replaced
  • Kafka 4.0 removes the `log.message.timestamp.difference.max.ms` service-level setting.
  • Use `log.message.timestamp.before_max_ms` and `log.message.timestamp.after_max_ms` instead.
  • For details, see KIP-937.
• KRaft-only clusters
  • Kafka 4.0 requires all clusters to run in KRaft mode. ZooKeeper is no longer supported for metadata management.
  • If your service still uses ZooKeeper, upgrade to Kafka 3.9 and migrate to KRaft before moving to Kafka 4.0.
  • For details, see KRaft mode in Aiven for Apache Kafka.

See our Aiven for OpenSearch documentation to learn how to use:

• Custom repositories
• Snapshots

Learn more in the Quick Connect documentation for Aiven for Apache Kafka.

What’s in

This release contains infrastructure and maintenance improvements and fixes for both OpenSearch and OpenSearch Dashboards. It also introduces JWT as an authentication type for OpenSearch Dashboards.

What’s out

The Learning to Rank plugin has been excluded from this upgrade and will be introduced with a future release. Stay tuned!

For the complete list of the 2.19.3 changes, see OpenSearch Release Notes 2.19.3.

New

• Added support for OIDC authentication and authorization, enabling secure access with OAuth-compatible providers.

Improved

• Re-added the karapace_version field to the Schema Registry health check endpoint for easier monitoring and version tracking.

Fixed

• Addressed issues with producer acknowledgements (acks) to improve message delivery reliability.
• Replaced the deprecated cgi.parse_header with updated header parsing that reuses a single function and correctly parses parameters.

For more information, see the Karapace 5.0.1 release notes.

For more information, see the OpenSearch sink connector release notes.

Aiven for AlloyDB Omni is being discontinued on the Aiven Platform:

• End of availability (EOA): After 5 September 2025, you can no longer create new services. Existing services will continue to operate until the end of life (EOL) date, but you will not be able to change service plans.
• End of life (EOL): On 5 December 2025, all active services will be decommissioned and deleted. This action is irreversible, and all data from these services will become inaccessible.

What you can do

After the EOL date, all active Aiven for AlloyDB Omni services will be deleted, rendering your data inaccessible. Plan and execute a timely migration to avoid data loss and service disruptions.

We recommend migrating your data to our alternative services: Aiven for PostgreSQL®, Aiven for ClickHouse®, or Aiven for MySQL® depending on your Aiven for AlloyDB Omni usage (analytics or transactional workloads for example).

For more information, see our EOL documentation.

For details, see the Karapace 4.1.3 release.

See how to enable and configure Aiven for ClickHouse® - Aiven for Kafka® integration.

• Added support for ingesting into Iceberg tables.
• Improved Snowpipe Streaming, including new defaults and removal of double-buffered mode.
• Fixed a critical issue in schema evolution that could skip records during schema changes.
• Recommended upgrade if using `snowflake.enable.schematization` or `snowflake.topic2table.map`.
• Applied multiple security fixes, including CVE-2025-48924.

For the complete list of changes, see the release notes.

You can now restore Aiven for OpenSearch® snapshots using the Aiven console or API.

How to get access

This feature is currently in limited availability (LA). To gain access and start using it, reach out to Aiven Support or your account team.

How to use it

Once enabled, you can find instructions on how to restore Aiven for OpenSearch® snapshots in our documentation.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

See all Azure cloud regions, where the Azure Private Link early availability feature is supported on Aiven.

See the release notes for all TimescaleDB releases.

• Improved processing of Kafka.Decimal and Debezium VariableScaleDecimal types.
• When useCredentialsProjectId is set to true, the connector now uses the project defined in the configuration.
• Retained the original topic, partition, and offset in Kafka metadata records.
• Added warning logs for lookup retries.

For more information, see the GitHub release notes.

PostgreSQL Anonymizer enables you to:

• Comply with privacy regulations: By masking or replacing sensitive data (PII).
• Safely use data in non-production environments: You can work with realistic, yet anonymized, datasets.
• Apply various masking strategies: Including static (permanent), dynamic (real-time, role-based), and anonymous data dumps.
• Define masking rules directly in SQL: Making it a declarative and integral part of the database schema.
• Generate fake but realistic data: For more effective testing and development.

Learn how to use postgresql_anonymizer on Aiven in Extensions on Aiven for PostgreSQL® > Utilities.

Key benefits

• Reduced DDL lock times: For large databases, full backups can cause prolonged DDL (Data Definition Language) locks. Incremental backups significantly reduce this lock duration, especially when there haven't been many changes.
• Efficient storage: Databases with fewer changes will benefit from less storage consumption, as incremental backups only store the differences.

Aiven for MySQL® incremental backups is a limited availability feature. For access, contact your account team. For more information, see our documentation.

• Deprecated the GCS-based batch loading feature. This will be removed in a future release.
• Deprecated the legacy partition decorator syntax. Scheduled for removal in version 3.0.0.
• Added user-agent and version headers to outgoing requests to improve observability.
• Improved reliability by retrying getTable() calls with backoff before failure.
• Added optional support for converting Debezium VariableScaleDecimal types to BigDecimal, preserving precision for PostgreSQL NUMERIC fields.
• Added a configuration option to extract the GCP project ID from the service account key file, useful in multi-project setups.

For more information, refer to the GitHub release notes for the BigQuery connector for Apache Kafka.

For details on PostgreSQL version support lifecycles on the Aiven Platform, consult our Aiven for PostgreSQL end-of-life documentation.

This version includes:

• Dependency updates and security fixes, including a fix for CVE-2025-46762.
• Fix for a race condition that could cause connector startup failures when using slow JDBC catalogs.
• Improved support for STS AssumeRole authentication with cloud object storage.

For more details, see the Apache Iceberg 1.9.1 release notes.
To learn how to use the connector, see the Iceberg sink connector documentation.

Version 1.9 is not compatible with Apache Kafka Connect 3.9. A patched version is now available for compatibility. For more information, see the MongoDB Kafka Source Connector documentation.

Features

• #8081 Use JSON error code for job configuration parsing
• #8100 Support splitting compressed chunks
• #8131 Add policy to process hypertable invalidations
• #8141 Add function to process hypertable invalidations
• #8165 Reindex recompressed chunks in compression policy
• #8178 Add columnstore option to CREATE TABLE WITH
• #8179 Implement direct DELETE on non-segmentby columns
• #8182 Cache information for repeated upserts into the same compressed chunk
• #8187 Allow concurrent Continuous Aggregate refreshes
• #8191 Add option to not process hypertable invalidations
• #8196 Show deprecation warning for TAM
• #8208 Use NULL compression for bool batches with all null values like the other compression algorithms
• #8223 Support for attach/detach chunk
• #8265 Set incremental Continuous Aggregate refresh policy on by default
• #8274 Allow creating concurrent continuous aggregate refresh policies
• #8314 Add support for timescaledb_lake in loader
• #8209 Add experimental support for Direct Compress of COPY
• #8341 Allow quick migration from hypercore TAM to (columnstore) heap

Bug fixes

• #8153 Restoring a database having NULL compressed data
• #8164 Check columns when creating new chunk from table
• #8294 The "vectorized predicate called for a null value" error for WHERE conditions like x = any(null::int[]).
• #8307 Fix missing catalog entries for bool and null compression in fresh installations
• #8323 Fix DML issue with expression indexes and BHS

GUCs

• enable_direct_compress_copy: Enable experimental support for direct compression during COPY, default: off
• enable_direct_compress_copy_sort_batches: Enable batch sorting during direct compress COPY, default: on
• enable_direct_compress_copy_client_sorted: Correct handling of data sorting by the user is required for this option, default: off

Deprecation

• Hypercore access method

See the release notes for all TimescaleDB releases.

Use the generator to:

• Choose from sample scenarios, such as logistics, user activity, or metrics
• Start streaming data within seconds of service creation
• View messages and schema details in the Aiven Console
• Stop the stream from the browser tab where it was started

Only one sample stream can run per service at a time. This feature is available at no additional cost and does not use service credits.
For more information, see Stream sample data from the Aiven Console.

Key changes

• Added support for mutual TLS (mTLS) to improve security with client and server certificate validation.
• Migrated to FastAPI and Uvicorn for improved performance and scalability.
• Integrated OpenTelemetry for standardized telemetry and tracing.
• Improved REST-proxy support for reading messages with an Avro value and a string key.
• Fixed consumer commit handling and script entrypoints.

Breaking change

• Changed the response format of the health check endpoint (/_health). Some fields were removed, existing attributes are now nested under status {}, and a new healthy flag was added.

For more details, see the Karapace 5.0.0 release notes.

Enable AWS PrivateLink for your BYOC services to ensure secure and private communication.

This version of the Aiven Operator also supports Diskless Topics for Aiven for Apache Kafka® services. For BYOC services that have access to the this feature, the config.inkless_enable field in the KafkaTopic resource enables diskless.

For details on the updates and deprecations, check out the complete Aiven Operator changelog.

• Connector class names and converter package names have been standardized. Update configurations if they reference “com.datamountaineer” packaged class names.
• StreamReactor 9.x incorporates class name changes introduced in version 6.0.0 by Lenses. These changes make version 9.x incompatible with version 4.2.0 previously offered by Aiven.
• When upgrading from 4.2.0 to 9.x, recreate the connectors using the updated class names, as existing ones will not work with the new version. For details, see All connectors (6.x).
• To maintain compatibility and safe migration to 9.0.1, first pin the connector version to 4.2.0. For instructions, see Manage connector versions.

For more information, see the Stream Reactor release notes. For a list of supported Stream Reactor connectors in Aiven, see the list of connector plugins.

The Exoscale cloud support is a limited availability feature. For more information or access, contact your account team.

See also

• Other supported Exoscale regions
• All supported cloud providers and regions

Key considerations

• Migration from ZooKeeper to KRaft for existing services is not yet available. Support will be added later.
• Kafka 3.8 support is extended by one year to support transition planning.
• Startup-4 replaces Startup-2 plans in Apache Kafka 3.9 and later. All feature restrictions from Startup-2 also apply to Startup-4, including Datadog restrictions.
• Some controller metrics are no longer available in KRaft mode.
• Existing clients and integrations such as Kafka Connect, MirrorMaker 2, and Karapace continue to work without changes.

For details, see KRaft mode and Transitioning to KRaft.

This release also includes fixes for the following:

• project_vpc_id can now be set to null. When upgrading, verify your service VPC settings are correct as services without this field will be deployed to the public internet.
• The aiven_clickhouse_grant resource no longer attempts to manage or interfere with privileges granted on ClickHouse named collections, preventing errors when grants are managed externally.

In addition, several new fields have been added or updated. For more details on these fixes and updates, view the complete Aiven Provider for Terraform changelog.

New features

• Optimized plans: Introduced enhanced plans across AWS, Google Cloud, and Microsoft Azure, offering improved price-performance ratios and greater flexibility. You can now double compute power for the same plan or reduce costs by up to 30% for the same compute.
• Dynamic Disk Sizing (DDS): Enabled independent scaling of local storage, allowing you to pay only for the storage they need and avoid overprovisioning.

Improvements

• Utilizing more modern instance types, including ARM chipsets where available, for performance and cost improvements.
• Expanded plan options within startup and business tiers (new startup & business 8, and 128 plans) for simplified resource management and optimized costs.

Deprecation

• Older plans for Aiven for ClickHouse will be retired on August 29, 2025.

Learn more about the introduced enhancements and check out our FAQs in Your ClickHouse, optimized: Experience more flexible, price-performant plans.

Features

• #8145 Log only if compression ratio warnings are enabled

Bug fixes

• #7292 Intermittent "could not find pathkey item to sort" error when grouping or ordering by a time_bucket of an equality join variable
• #8126 Allow setting bgw_log_level to FATAL and ERROR
• #8151 Treat null equal to null for merged Continuous Aggregate refresh
• #8153 Restoring a database having NULL compressed data
• #8162 Fix setting compress_chunk_interval on continuous aggregates
• #8163 Fix gapfill crash with locf NULL values treated as missing
• #8171 Disable decompression limit during continuous aggregate refresh

See the release notes for all TimescaleDB releases.

Use this connector to ingest data from Azure Blob Storage containers into Kafka topics for real-time processing, analytics, or disaster recovery. The connector supports multiple input formats, including bytes, jsonl, avro, and parquet. You can configure advanced options such as schema registry integration, file compression, topic overrides, and fault-tolerant recovery.

For more information, see the Azure Blob source connector documentation.

What’s new:

• Debezium 3.1.0 is now the default version for new connectors.
• Older versions, including 1.9.7, remain available through version pinning.
• Connectors are not upgraded automatically. To upgrade, set the version using the plugin_versions configuration.
• Debezium versions 2.0 and later do not support the wal2json replication format. If you use wal2json with PostgreSQL, stay on version 1.9.7 or migrate to a supported format such as pgoutput.

To upgrade:
Set the desired version using the plugin_versions property. Test any changes in a staging environment before applying them in production.

For more information, see:

• Manage connector versions
• Debezium 3.1.0 release notes

You can now conveniently create and manage OpenSearch snapshot repositories directly from the Aiven Console.

How to get access

This feature is currently in limited availability (LA). To gain access and start using it, get in touch with Aiven Support or your account team.

Once enabled, you can find the comprehensive setup and usage details in the custom snapshot repositories guide in our documentation.

You can now manage snapshots with the standard OpenSearch snapshot API, with no Aiven-specific endpoints required. This means a more seamless and familiar experience for you, with full control over:

• Repository operations: register, verify, get, delete, cleanup
• Snapshot operations: create, get, status, delete, restore, clone

How to get access

This feature is currently in limited availability (LA). To gain access and start using the native OpenSearch snapshot API, reach out to Aiven Support or your account team.

Once enabled, you can find the complete setup guide and instructions in our documentation.

• Austria, Vienna: AT-VIE-1
• Austria, Vienna: AT-VIE-2
• Bulgaria, Sofia: BG-SOF-1
• Germany, Frankfurt: DE-FRA-1
• Germany, Munich: DE-MUC-1
• Switzerland, Zurich: CH-DK-2
• Switzerland, Geneva: CH-GVA-2

The Exoscale cloud support is a limited availability feature. For more information or access, contact your account team.

Check all cloud regions available on Aiven in the documentation.

Key features:

• Use `key.fields` to specify fields to extract from the key.
• Use `value.fields` to rename or map key fields into the value.
• Supports copying the entire key using *.

For more information, see the GitHub README.

• Added experimental Single Message Transforms (SMTs) for Debezium and AWS DMS to support CDC use cases.
• Improved security and stability through upstream Apache Iceberg updates.

For more information, see the Iceberg sink connector documentation and the Apache Iceberg 1.9.0 release notes.

To get started, see Create an Iceberg sink connector.

To learn more about using Private Service Connect with Aiven-managed services, see the documentation.

As of 15 May 2025, the storage allocated to new and existing free plans for Aiven for PostgreSQL® and Aiven for MySQL services has been adjusted from 5GB to 1GB.

No action is needed regarding your stored data and the service connection details will stay the same. You will still have more than 50 percent of your disk space available. For users who require more storage, Aiven has hobbyist plans for PostgreSQL and MySQL.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

This release contains the following bug fixes on top of 2.19.2:

• #7893 Don't capture hard errors in with-clause parser
• #7903 Don't capture hard errors for old cagg format
• #7912 Don't capture errors estimating time max spread
• #7910 Fix not using SkipScan over one chunk
• #7913 Allow TAM chunk creation as non-owner
• #7935 Fix TAM segfault on DELETE using segmentby column
• #7954 Allow scheduler restarts to be disabled
• #7964 Crash when grouping by multiple columns of a compressed table, one of which is a UUID segmentby column.

See the release notes for all TimescaleDB releases.

This release contains the following changes on top of 2.19.1:

• Feature #7923: Add a GUC to set a compression batch size limit
• GUC: compression_batch_size_limit: set batch size limit, default: 1000
• Bug fix #7911: Don't create a Hypertable for published tables
• Bug fix #7902: Fix crash in VectorAgg plan code

See the release notes for all TimescaleDB releases.

Key highlights:

• Built-in generative AI: Build and deploy AI applications directly on your operational data using SQL. Perform vector similarity searches and leverage advanced AI models seamlessly.
• High performance: Optimized for transactional, analytical, and vector search workloads, enabling real-time insights and efficient AI-powered solutions.
• Multi-cloud flexibility: Deploy across AWS, Google Cloud, or Azure with centralized management through the Aiven platform.
• Enterprise-grade security: Enjoy end-to-end encryption, private networking, and role-based access control for peace of mind.
• Seamless scalability: Scale your database, upgrade versions, or create read replicas without downtime.

Aiven for AlloyDB Omni empowers businesses to harness the power of AI and advanced database technology with ease, scalability, and security. Start building today!

Learn more about Aiven for AlloyDB Omni.

Learn more about Aiven for ClickHouse backups in the documentation.

Check all supported Google Cloud regions in the documentation.

Learn more about Aiven’s organization VPCs and supported peering scenarios.

pgvectorscale enhances the pgvector extension, allowing you to efficiently store, index, and search high-dimensional vector embeddings for AI/ML applications. It offers automatic scaling and optimized performance for similarity search. pgvectorscale brings improved query speed, cost-effective scalability, and an SQL-based interface for AI-powered search and recommendation systems.

For more information, see the pgvectorscale documentation.

Learn how to configure your backup schedule in the documentation.

See the documentation to learn how to manage Aiven for ClickHouse integrations with the new console features.

For more information, see:

• Kafka 3.9 release notes
• KRaft documentation
  

• Change how client.Error is initialized.

For more details, see the Aiven CLI release notes.

Check all supported Google Cloud regions in the documentation.

Check all supported Google Cloud regions in the documentation.

See all the supported OCI regions in the documentation.

The OCI regions support is a limited availability feature. For more information or access, contact your account team.

See the documentation for all cloud providers and regions supported on the Aiven Platform.

Additionally, modifying the deprecated aiven_organization_group_project and aiven_project_user resources is now restricted as these have been replaced by the aiven_organization_permission resource. Follow the migration guide to move your existing resources to the permission resource.

For more details on fixes and updates, view the complete Aiven Provider for Terraform changelog.

Highlights

• JOIN improvements, including disk spillover for cross joins
• Ability to iterate over archive container files in S3
• Recursive CTEs
• New table engines and related functions:
  • loop allows to return query results in an infinite loop.
  • fuzzQuery allows to randomly mutate query strings.
• New functions such as generateUUIDv7, generateSnowflakeID (along with conversion functions for date times such as snowflakeIDToDateTime), and base64 functions base64URLEncode & base64URLDecode
• New numpy array output format NPy
• New system tables:
  • detached_tables to keep track of currently detached tables
  • error_log to persist error history
  • text_log for logging entries
• More information added to the existing query_log and part_log system tables
• Lightweight deletes improvements, along with a new lightweight_deletes_sync setting to control their behavior
• Filesystem cache improvements through contention minimization and usage optimization
• Performance improvements

Data types

• JSON, Dynamic, Variant are now supported as experimental column data types and will be made widely available during the next version upgrade.
• The Object('json') data type and its JSON alias
  • After the upgrade, Object('json') and JSON are two different data types. Start using Object('json') now.
  • Migration from the Object('json') type to the JSON type will be needed, and Aiven will provide guidance to support you through the process.

Query analyzer

The new query analyzer has graduated from experimental to beta. It will be enabled for all services during the next version release. We’d love to already hear and consider your feedback on using it. To try out the new query analyzer, use the following setting:

SELECT *

FROM table

SETTINGS enable_analyzer=1;

Table engines

• Experimental TimeSeries table engine has been added, allowing to store metrics using the Prometheus protocol. The engine is not production-ready and hasn’t been enabled yet.
• S3Queue storage engine is undergoing testing and stabilization. The engine is not production-ready and hasn’t been enabled yet.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

• Add MySQL client support for the avn service cli command line.
• Allow privatelink refresh for AWS.
• Add get_service_integration_endpoint.

For more details, see the Aiven CLI release notes.

Existing customers using Startup-2 with Datadog integration can continue to create Startup-2 plans that include Datadog and use their existing services without upgrading to a higher plan. To continue using Datadog, all other customers must choose a Business-4 plan or higher.

For more details, see the Datadog integration documentation.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

An organization VPC allows you to create a single VPC peering connection to multiple Aiven project within the organization. Beyond that, organization VPCs bring simplified connectivity and networking, easier operations and maintenance, improved scalability and flexibility, and reduced complexity.

Learn more about Aiven’s organization VPCs and supported peering scenarios.

In this release, the Aiven for Cassandra® and Aiven for Caching (formerly Aiven for Redis®*) resources were deprecated. These services are approaching their end of life on the Aiven Platform.

For details on this release and deprecations, check out the complete Aiven Kubernetes Operator changelog.

End of availability: 30 November 2025

After this date, you can no longer create new Aiven for Apache Cassandra services. Existing services continue to operate until the end of life (EOL) date.

End of life: 31 December 2025

After 31 December 2025, all active Aiven for Apache Cassandra services are powered off and deleted, making data from these services inaccessible.

To ensure uninterrupted service with no data loss, complete your migration out of Aiven for Apache Cassandra before 31 December 2025. For further assistance, contact your account team.

• ap-osaka-1: Japan Central: Osaka
• ap-seoul-1: South Korea Central: Seoul
• ap-singapore-1: Singapore: Singapore
• ap-tokyo-1: Japan East: Tokyo

See all the supported OCI regions in the documentation.

The OCI regions support is a limited availability feature. For more information or access, contact your account team.

Highlights

• Performance improvements
  • Asynchronous I/O Threading: Enables parallel processing of commands and I/O operations, maximizing throughput and minimizing bottlenecks.
  • Intelligent Core Utilization: Distributes I/O tasks across multiple cores based on realtime usage, reducing idle time and improving energy efficiency.
  • Command Batching: Optimizes memory access patterns by prefetching frequently accessed data to minimize CPU cache misses, reducing memory accesses required for dictionary operations.
• Efficiency improvements
  • Reduction of the memory overhead of keys, allowing you to store more data without any application changes: Keys are now embedded in the main dictionary, eliminating separate key pointers and significantly reducing memory overhead. This results in a 9-10% reduction in overall memory usage for scenarios with 16-byte keys and 8 or 16-byte values, along with performance improvements.
  • Improved Pub/Sub Efficiency: Lightweight cluster messages streamline communication and reduce overhead for faster, more efficient Pub/Sub operations.
• Replication improvements
  • Reduced Memory Load: By streaming replication data to the replica during the full sync, the primary node experiences significantly less memory pressure. The replica now manages the Client Output Buffer (COB) tracking, reducing the likelihood of COB overruns and enabling larger COB sizes on the replica side.
  • Reduced Parent Process Load: A dedicated connection for RDB transfer frees the primary's parent process from handling this data, allowing it to focus on client queries and improving overall responsiveness.

See the release notes for more details.

The following fields have also been added to the aiven_opensearch resource and data source in this release:

• opensearch_user_config.opensearch.cluster_routing_allocation_balance_prefer_primary: When set to true, OpenSearch attempts to evenly distribute the primary shards between the cluster nodes.
• opensearch_user_config.opensearch.segrep: When set to true, it enables segment replication backpressure, a shard-level rejection mechanism that dynamically rejects indexing requests as replica shards in your cluster fall behind primary shards.

For more details on fixes and updates, view the complete Aiven Provider for Terraform changelog.

It delivers actionable insights and optimization recommendations, including query rewrites, index creation or removal suggestions, and more—helping you maintain peak database performance.

Learn more in our documentation.

Key highlights of this feature are:

• Roles and permissions at different levels: Grant access at the organization, unit and project level.
• Tailored access rules: Assign permissions to individual users or groups, ensuring users only have access they need to do their jobs.
• Streamlined workflows: There’s a single Aiven API and Aiven Terraform Provider resource to grant roles and permissions following the resource, principal, and action methodology.

Learn more about the new roles and permissions and get started by granting access to your organization, units, or projects.

Three new beta Aiven for AlloyDB Omni resources and data sources were added: aiven_alloydbomni, aiven_alloydbomni_database, and aiven_alloydbomni_user. Aiven for AlloyDB Omni is a high-performance PostgreSQL-compatible database featuring a columnar engine and built-in AI capabilities.

In addition, the aiven_service_integration resource now supports setting source and destination projects, enabling the creation of integrations between services in different projects.

For more details on fixes and updates, view the complete Aiven Provider for Terraform changelog.

• Kafka-native ACLs: Enable fine-grained control for Clusters, Topics, Groups, and TransactionalId resources, with support for ALLOW and DENY rules.
• Aiven ACLs: Simplify access control with predefined permissions and wildcard support for flexible configuration.

Kafka-native ACLs can be used in parallel with Aiven ACLs, with no requirement to migrate or choose one over the other. If only Kafka-native ACLs are needed, Aiven ACLs can be removed from the service. Both ACL types are configurable through the Aiven Console, Aiven CLI, Aiven API, and Aiven Provider for Terraform. To use Kafka-native ACLs, apply the required service maintenance updates.

To learn more, see Access Control Lists in Aiven for Apache Kafka®.

This version brings key capabilities and performance improvements:

• Enhanced query and vacuum performance, such as faster sequential reads and reduced vacuum memory usage
• Logical replication improvements, such as failover control and preserving slots during upgrades
• Developer features, such as constructors, identity functions, and the JSON_TABLE() function
• Improved bulk loading with the new COPY option ON_ERROR ignore
• Strengthened security with direct TLS handshake options

For more information on what PostgreSQL version 17 introduces, see PostgreSQL 17 release notes.

• Added support for Kafka-native ACL commands in Aiven for Apache Kafka.
• Updated Aiven for Apache Flink commands to ignore the current project.

For more details, see the Aiven CLI release notes.

Other additions include:

• aiven_opensearch resource field opensearch_user_config.opensearch.search_insights_top_queries
• aiven_thanos resource field thanos_user_config.private_access
• aiven_thanos resource field thanos_user_config.privatelink_access
• aiven_opensearch datasource field opensearch_user_config.opensearch.search_insights_top_queries
• aiven_thanos data source field thanos_user_config.private_access
• aiven_thanos data source field thanos_user_config.privatelink_access
• aiven_service_integration_endpoint resource field datadog_user_config.extra_tags_prefix
• aiven_service_integration_endpoint datasource field datadog_user_config.extra_tags_prefix

For more details, view the complete Aiven Provider for Terraform changelog.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

Learn more in our documentation.

For more information, see Advanced parameters.

Other additions in this release include:

• The ClickHouse field userConfig.recovery_basebackup_name for the backup to restore in a forked service.
• The Grafana field userConfig.auth_generic_oauth.use_refresh_token to refresh tokens and check access token expiration.
• The Kafka field userConfig.schema_registry_config.retriable_errors_silenced to silence errors that can be retried or custom errors you specify.
• The Kafka field userConfig.schema_registry_config.schema_reader_strict_mode to configure the Karapace schema-registry service to shutdown when there are invalid schema records in the _schemas topic.
• The OpenSearch fields userConfig.azure_migration.indices,  userConfig.gcs_migration.indices, and  userConfig.s3_migration.indices to define a list of indices to restore from the snapshot. Multi-index syntax is supported.

The following are deprecated:

• PostgreSQL field userConfig.additional_backup_regions
• Grafana field userConfig.alerting_enabled

For a full list of the changes and deprecations, check out the Aiven Operator changelog.

Refer to TimescaleDB release notes for all the changes introduced in this release.

Bug fixes

• #7384: Fix negative bitmapset member not allowed and performance degradation on queries to compressed tables with the ORDER BY clause matching the order of the compressed data
• #7388: Use-after-free in vectorized grouping by segmentby columns

Additionally, starting with this release, the changelog will now highlight every schema change. This should make it easier to track and understand the updates Aiven is delivering with the Terraform provider.

For more information on Aiven Provider for Terraform, view the documentation.

• avn byoc tags list
• avn byoc tags update --add-tag KEY=VALUE --remove-tag KEY
• avn byoc tags replace --tag KEY=VALUE

These tags are attached to BYOC environments and cascaded to bastion nodes in private deployment models.

For more information, see Aiven CLI release notes.

Refer to TimescaleDB release notes for all the changes introduced in this release.

Key new features

• Full PostgreSQL 17 support for all existing features:
  TimescaleDB 2.17.1 is available for PostgreSQL 14, 15, 16, and 17.
• Significant performance improvements for continuous aggregate policies:
  Continuous aggregate refresh is now using merge instead of deleting old materialized data and re-inserting. This update can decrease dramatically the amount of data that must be written on the continuous aggregate in the presence of a small number of changes, reduce the i/o cost of refreshing a continuous aggregate, and generate fewer Write-Ahead Logs (WAL). Overall, continuous aggregate policies will be more lightweight, use less system resources, and complete faster.
• Increased performance for real-time analytical queries over compressed hypertables:
  Introducing additional Single Instruction, Multiple Data (SIMD) vectorization optimization to our engine by supporting vectorized execution for queries that group by using the segment_by column(s) and aggregate using the basic aggregate functions (sum, count, avg, min, max)
• Improved performance of deletes on compressed hypertables when a large amount of data is affected:
  Speeding up operations that delete whole segments by skipping the decompression step (enabled for all deletes that filter by segment_by columns)

PostgreSQL 14 deprecation announcement

PostgreSQL 14 will be supported until April 2025. Closer to that time, there will be an announcement of the specific version of TimescaleDB in which PostgreSQL 14 support will not be included going forward.

New features

• #7360 Add chunk skipping GUC
• #6882: Allow delete of full segments on compressed chunks without decompression.
• #7033: Use merge statement on continuous aggregates refresh.
• #7126: Add functions to show the compression information.
• #7147: Vectorize partial aggregation for sum(int4) with grouping on segment by columns.
• #7204: Track additional extensions in telemetry.
• #7207: Refactor the decompress_batches_scan functions for easier maintenance.
• #7209: Add a function to drop the osm chunk.
• #7275: Add support for the returning clause for merge.
• #7200: Vectorize common aggregate functions like min, max, sum, avg, stddev, variance for compressed columns of arithmetic types, when there is grouping on segment by columns or no grouping.

Bug fixes

• #7335 Change log level used in compression
• #7342 Fix collation for in-memory tuple filtering
• #7187: Fix the string literal length for the compressed_data_info function.
• #7191: Fix creating default indexes on chunks when migrating the data.
• #7195: Fix the segment by and order by checks when dropping a column from a compressed hypertable.
• #7201: Use the generic extension description when building apt and rpm loader packages.
• #7227: Add an index to the compression_chunk_size catalog table.
• #7229: Fix the foreign key constraints where the index and the constraint column order are different.
• #7230: Do not propagate the foreign key constraints to the osm chunk.
• #7234: Release the cache after accessing the cache entry.
• #7258: Force English in the pg_config command executed by cmake to avoid the unexpected building errors.
• #7270: Fix the memory leak in compressed DML batch filtering.
• #7286: Fix the index column check while searching for the index.
• #7290: Add check for null offset for continuous aggregates built on top of continuous aggregates.
• #7301: Make foreign key behavior for hypertables consistent.
• #7318: Fix chunk skipping range filtering.
• #7320: Set the license specific extension comment in the install script.

For further assistance, contact your Aiven sales team.

• Systemd journal logging: Added support for logging to the systemd journal for easier log management.
• Schema compatibility checks: Introduced a new messages field in the compatibility API, which provides more precise explanations of schema issues.

To be enabled later:

• Protobuf formatter: This release includes preparation for the Protobuf formatter, which will be available in a future update. It will provide more flexibility when handling Protobuf schemas.

For more details, see the Karapace 4.0.0 release notes.

Key considerations:

• Once enabled, tiered storage cannot be disabled.
• Shutting down the service will result in permanent loss of data stored in tiered storage.
• Tiered storage is available only on AWS, Google Cloud, and Microsoft Azure.

For more information, see Get started with tiered storage and review the existing limitations for tiered storage.

• New string similarity functions: arrayRandomSample, editDistance, dramerauLevenshteinDistance, jaroSimilarity, and more
• Other new functions: arrayFold, groupArrayIntersect, largestTriangleThreeBuckets, seriesPeriodDetectFFT, and more
• Space-filling curves (such as mortonEncode) can be used for primary key, which is useful if there are two separate expressions (rather than just one) that are often filtered by.
• Automatic detection of a file format (for example, Parquet) for files with no extension
• ALTER materialized_view MODIFY query
• Support for NPY file format for better NumPy interoperability
• GCD compression codec for integers
• PASTE JOIN: create a table by joining all columns from a left subquery to all columns from a right subquery in the order of their respective rows
• Vectorized functions cosineDistance, dotProduct, and L2Distance for better performance and additional ClickHouse vector database capabilities
• Adaptive asynchronous INSERTs: adjust flush timeout based on frequency of INSERTSs
• Performance improvements

ClickHouse 24.3 brings changes to defaults of a few settings related to JSON, Parquet, ORC, and Arrow input/output formats:

• input_format_json_try_infer_numbers_from_strings: true → false
• output_format_parquet_string_as_string: false → true
• output_format_arrow_string_as_string: false → true
• output_format_orc_string_as_string: false → true

If you’re using these settings, append a SETTINGS section to the end of queries while upgrading, for example:

SELECT *
FROM table
FORMAT Parquet
SETTINGS output_format_parquet_string_as_string=0;

For more details on this release, see the OpenSearch Release Notes 2.17.1.

To get started, see Create an Azure Blob Storage sink connector documentation.

Key features:

• Centralized monitoring across multiple Prometheus servers
• Unlimited retention with scalable object storage
• Efficient query performance with downsampling and compacting
• Fully managed with optimized Thanos components for simplified operations

Aiven for Metrics allows you to focus on extracting insights and making better decisions from your data. For more details, see the Aiven for Metrics documentation.

Key features:

• PostgreSQL compatibility
• Multi-cloud flexibility
• Lightning-fast analytics
• Managed service simplicity
• Accelerated AI adoption
• Enterprise-grade security

Key use cases:

• Data-intensive applications with complex workloads
• High-performance transactional workloads
• Multi-cloud deployments
• High-traffic web applications
• Real-time analytics and dashboards
• Machine learning and AI applications

Aiven for AlloyDB Omni is in the early availability stage. You can easily share feedback on your Aiven for AlloyDB Omni experience in the Aiven Console.

See the Aiven for AlloyDB Omni documentation for more information.

• Aiven for AlloyDB Omni
• Aiven for Valkey ACL commands

For more information, see Aiven CLI release notes.

To learn more about using PSC with Aiven-managed services, see the documentation.

For more information on Aiven Provider for Terraform, check out the documentation.

• Fixed stage file deletion during Snowpipe ingestion to ensure proper cleanup of temporary files.
• Added a timestamp torecord_metadata to track when data is pushed to Snowflake. This field is optional and does not affect existing integrations.
• Included improvements from versions 2.2.1 and 2.2.2 for better performance and stability.

For more information, see the release notes.

To enable unified alerting, update the Advanced configuration parameters in the Aiven for Grafana service settings within the Aiven console:

• Set alerting_enabled to disabled.
• Set unified_alerting_enabled to enabled.

For more details, see the Grafana alert migration guide.

Removed plugins
The following plugins have been removed due to upstream deprecations:

• agenty-flowcharting-panel
• blackmirror1-singlestat-math-panel
• flant-statusmap-panel
• grafana-worldmap-panel
• natel-discrete-panel
• petrslavotinek-carpetplot-panel
• ryantxu-ajax-panel
• vonage-status-panel
• grafana-piechart-panel
• grafana-simple-json-datasource
• xginn8-pagerduty-datasource
• doitintl-bigquery-datasource

To view the list of supported plugins, see Aiven documentation on Grafana plugins.

For any assistance, contact Aiven Support via the Aiven Console or by email at support@aiven.io.

• Schema reader behavior: Karapace can be configured to shut down if it finds corrupt entries in the _schemas topic. By default, Karapace skips these entries. When configured to shut down, any error while consuming data from the _schemas topic will cause Karapace to stop.
• Shutdown metrics: Karapace now includes a metric to monitor how often it shuts down. This helps you track shutdown events more effectively.

For a complete list of changes, see the Karapace release notes.

Aiven for ClickHouse tiered storage is supported on the following cloud platforms: Microsoft Azure, AWS, Google Cloud.

Learn how to enable and configure Aiven for ClickHouse® tiered storage in our documentation.

• prefix_path_depth attribtue to kafka_user_config.kafka_connect_secret_providers.vault
• backup_hour and backup_minute to redis_user_config and valkey_user_config
• user configuration options for external_s3, external_clickhouse, and external_mysql integration endpoint types

This releae also increases the polling timeout for aiven_kafka_connector. For more details on other fixes and updates, view the complete Aiven Provider for Terraform changelog.

Refer to TimescaleDB release notes for 2.16.0 and TimescaleDB release notes for 2.16.1 for all the changes introduced in these release.

Key new features

• Performance-focused optimizations for data manipulation operations (DML) over compressed chunks: Improved UPSERT performance by more than 100x in some cases and more than 1000x in some update/delete scenarios.
• Ability to define chunk skipping indexes on non-partitioning columns of compressed hypertables: Extended chunk exclusion to use those skipping (sparse) indexes when queries filter on the relevant columns and prune chunks that don't include any relevant data for calculating the query response.
• For use cases that require foreign keys defined:
  • Ability to add foreign keys from regular tables towards hypertables
  • Removed some locks in the reverse direction that blocked access to referenced tables while compression was running.
• Extended Continuous Aggregates to support more types of analytical queries:
  • More types of JOINs supported
  • Additional equality operators on JOIN clauses supported
  • JOINs between multiple regular tables supported

More new features

• #6880: Added support for the array operators used for compressed DML batch filtering.
• #6895: Improved the compressed DML expression pushdown.
• #6897: Added support for replica identity on compressed hypertables.
• #6918: Removed support for PG13.
• #6920: Reworked compression activity WAL markers.
• #6989: Added support for foreign keys when converting plain tables to hypertables.
• #7020: Added support for the chunk column statistics tracking.
• #7048: Added an index scan for INSERT DML decompression.
• #7075: Reduced decompression on the compressed INSERT.
• #7101: Reduced decompressions for the compressed UPDATE/DELETE.
• #7108 Reduced decompressions for INSERTs with UNIQUE constraints.
• #7116 Used DELETE instead of TRUNCATE after compression.
• #7134 Refactored foreign key handling for compressed hypertables.
• #7161 Fixed mergejoin input data is out of order.

Improvements

• Improved query performance through chunk exclusion on compressed hypertables: You can now define chunk skipping indexes on compressed chunks for any column with one of the following integer data types: smallint, int, bigint, serial, bigserial, date, timestamp, timestamptz. After you call enable_chunk_skipping on a column, TimescaleDB tracks the min and max values for that column. TimescaleDB uses that information to exclude chunks for queries that filter on that column, and would not find any data in those chunks.
• Improved UPSERT performance on compressed hypertables: By using index scans to verify constraints during inserts on compressed chunks, TimescaleDB speeds up some ON CONFLICT clauses by more than 100x.
• Improved performance of UPDATEs, DELETEs, and INSERTs on compressed hypertables: By filtering data while accessing the compressed data and before decompressing, TimescaleDB has improved performance for UPDATEs and DELETEs on all types of compressed chunks, as well as INSERTs into compressed chunks with unique constraints. By signaling constraint violations without decompressing or decompressing only when matching records are found in the case of UPDATEs, DELETEs, and UPSERTs, TimescaleDB 2.16.0 speeds up those operations more than 1000x in some UPDATE/DELETE scenarios, and 10x for UPSERTs.
• Ability to add foreign keys from regular tables to hypertables with support for all types of cascading options; useful for hypertables that partition using sequential IDs and need to reference those IDs from other tables.
• Lower locking requirements during compression for hypertables with foreign keys: Advanced foreign key handling removes the need for locking referenced tables when new chunks are compressed. DML is no longer blocked on referenced tables while compression runs on a hypertable.
• Improved support for queries on Continuous Aggregates
  • INNER/LEFT and LATERAL JOINs are now supported.
  • Ability to join with multiple regular tables
  • Ability to have more than one equality operator on JOIN clauses

Bug fixes

• #7182 Fixed untier_chunk for hypertables with foreign keys.
• #6987 Fixed REASSIGN OWNED BY for background jobs.
• #7018: Fixed search_path quoting in the compression defaults function.
• #7046: Prevented locking for compressed tuples.
• #7055: Fixed the scankey for segment by columns, where the type constant is different to variable.
• #7064: Fixed the bug in the default order by calculation in compression.
• #7069: Fixed the index column name usage.
• #7074: Fixed the bug in the default segment by calculation in compression.

• AWS Secrets Manager
• HashiCorp Vault

For more information, see Aiven for Apache Kafka® Connect secret providers.

To learn more about using dictionaries in Aiven for ClickHouse, see the documentation.

Managed credentials integration in Aiven for ClickHouse is supported with the following data source types: PostgreSQL®, MySQL®, ClickHouse®, and Amazon S3®.

Learn how to enable and use Aiven for ClickHouse managed credentials integration in our documentation.

Improvements

• Improved the handling of MirrorCheckpointTask restarts to prevent permanent interruptions in offset translation (KAFKA-15905).

Bug fixes

• Fixed an issue where the auto.offset.reset=latest configuration did not function as expected, ensuring topics are read from the latest offset (KAFKA-13988).
• Addressed a potential null pointer exception (NPE) in MirrorCheckpointTask (KAFKA-15607).
• Resolved a problem where the first checkpoint was not emitted until the consumer group had fully caught up (KAFKA-16622).

For further details and updates, see the Apache Kafka release notes.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

• Schema name handling: Resolved issues with fully qualified paths versus simple name references in schemas.
• Process stability: Enhanced exception handling in forked processes to prevent timeouts.
• Version logging: Corrected version number formatting in schema_reader.py for accurate logging and easier troubleshooting.

Compatibility and updates

• Python 3.12 support: Added compatibility with Python 3.12.
• Confluent API update: Updated to the latest Confluent API specifications for continued compatibility and smoother integration with Confluent clients.

For a complete list of fixes and detailed information, see the Karapace release notes.

At the same time, we're introducing more BYOC enhancements:

• You can apply custom labels to both service VMs and shared BYOC infrastructure, for example VPCs.
• You can choose between public and private BYOC deployment models for their Aiven-managed services.

Learn more about BYOC in the documentation.

• Scalability: Independently scale storage and compute resources.
• Cost efficiency: Offload less critical data to more cost-effective remote storage.
• Operational speed: Boost performance and enable faster transitions between cloud providers.
• Granular control: Customize data management with per-topic storage settings.
• Infinite retention: Store data indefinitely without space constraints.

Learn more in our documentation on Tiered Storage in Aiven for Apache Kafka.

Enhancements:

• Exactly once delivery support (KIP-618): Ensures messages are delivered exactly once.
• Tombstone handling: Enables row delete functionality, allowing the connector to handle tombstone messages effectively.
• Primary key (PK) mode validation: Adds validation for PK mode against PK fields to ensure data integrity and correctness.

Bug fixes:

• Source task shutdown correction: Improved shutdown logic for source tasks to avoid memory leaks.

For more information, see release notes and documentation.

Refer to the TimescaleDB release notes for 2.15.3 to check all the changes introduced in this release, including deprecation announcements and operational instructions.

Bug fixes

• #7061: Fix the handling of multiple unique indexes in a compressed INSERT.
• #7080: Fix the corresponding equivalence member not found error.
• #7088: Fix the leaks in the DML functions.
• #7035: Fix the error when acquiring a tuple lock on the OSM chunks on the replica.
• #7091: Fix the ORDER BY/GROUP BY expression not found in targetlist error on PG16.

The aiven_project_user deprecation has also been removed in this release, meaning you can continue to use this resource.

For more details on fixes and updates, view the complete Aiven Provider for Terraform changelog.

To get started, set up Okta as an identity provider for your organization and configure SCIM user provisioning.

Refer to the TimescaleDB release notes for 2.15.2 to check all the changes introduced in this release, including deprecation announcements and operational instructions.

Key new features

• Continuous Aggregate now supports time_bucket with origin and/or offset.
• For hypertable compression:
  • Recommend optimized defaults for segment by and order by when configuring compression through analysis of table configuration and statistics.
  • Add planner support to check more kinds of WHERE conditions before decompression.
    This reduces the number of rows that have to be decompressed.
  • You can now use minmax sparse indexes when you compress columns with btree indexes.
  • Vectorize filters in the WHERE clause that contain text equality operators and LIKE expressions.

More new features

• #6382 Support for time_bucket with origin and offset in CAggs
• #6696 Improve the defaults for compression segment_by and order_by
• #6705 Add sparse minmax indexes for compressed columns that have uncompressed btree indexes
• #6754 Allow DROP CONSTRAINT on compressed hypertables
• #6767 Add metadata table _timestaledb_internal.bgw_job_stat_history for tracking job execution history
• #6798 Prevent the usage of the deprecated time_bucket_ng in the CAgg definition
• #6810 Add telemetry for access methods
• #6811 Remove the no longer relevant timescaledb.allow_install_without_preload GUC
• #6837 Add migration path for CAggs using time_bucket_ng
• #6865 Update the watermark when truncating a CAgg

Deprecation announcements

• You can no longer create continuous aggregates using time_bucket_ng. This feature will be completely removed in the next release. The best practice is to migrate your current continuous aggregates to the new form.
• This is the last release supporting PostgreSQL 13.

Bug fixes

• #6975 Fix sort pushdown for partially compressed chunks.
• #6976 Fix removal of metadata function and update script.
• #6978 Fix segfault in compress_chunk with a primary space partition.
• #6993 Disallow hash partitioning on primary column.
• #6617 Fix error in show_chunks
• #6621 Remove metadata when dropping chunks
• #6677 Fix snapshot usage in CAgg invalidation scanner
• #6698 Define meaning of 0 retries for jobs as no retries
• #6717 Fix handling of compressed tables with primary or unique index in COPY path
• #6726 Fix constify cagg_watermark using window function when querying a CAgg
• #6729 Fix NULL start value handling in CAgg refresh
• #6732 Fix CAgg migration with custom timezone / date format settings
• #6752 Remove custom autovacuum setting from compressed chunks
• #6770 Fix plantime chunk exclusion for OSM chunk
• #6789 Fix deletes with subqueries and compression
• #6796 Fix a crash involving a view on a hypertable
• #6797 Fix foreign key constraint handling on compressed hypertables
• #6816 Fix handling of chunks with no constraints
• #6820 Fix a crash when the ts_hypertable_insert_blocker was called directly
• #6849 Use non-orderby compressed metadata in compressed DML
• #6867 Clean up compression settings when deleting compressed CAgg
• #6869 Fix compressed DML with constraints of form value OP column
• #6870 Fix bool expression pushdown for queries on compressed chunks

To learn how to create and use SQL UDFs in Aiven for ClickHouse, see the documentation.

To learn more, see the documentation.

Features:

• Prometheus metrics and instrumentation: Added support for Prometheus metrics and instrumentation, enhancing observability and monitoring.

Bug fixes:

• ClientError handling: Implemented handling for aiohttp.ClientError with a 503 response status, improving error handling.
• TimeoutError handling: Added handling for asyncio.TimeoutError, to ensure smoother operation under timeout conditions.
• Invalid version handling: Fixed the issue where an invalid version error was raised for schema version 0. Schema versions greater than or equal to 0 are allowed, enhancing compatibility.

For a complete list of fixes and detailed information, see the Karapace release notes.

Bug fixes

• Data loss issue (KAFKA-16692): Fixed a critical issue that caused ADD_PARTITIONS_TO_TXN requests to fail during upgrades from Kafka 3.5 to 3.6. With this update, you no longer need to disable transaction.partition.verification.enable before upgrading.
• Tiered storage rebalances (KAFKA-16890): Fixed a bug affecting tiered storage rebalances, improving the performance and reliability of storage operations.

For a comprehensive list of all bug fixes and improvements, refer to the Apache Kafka 3.7.1 release notes.

More details on Apache Cassandra 4.1.5 in Apache Cassandra release notes.

Account Owners will transition to super admin

On 30 September 2024 the Account Owners team will be removed. Super admin have full access to organizations. The Account Owners and super admin are synced, so the removal of the Account Owners team will have no impact on existing permissions.

Teams will become read-only

From 4 November 2024 you won’t be able to create new teams or update existing ones. To simplify the move, we’ll also migrate your existing teams to groups.

Teams will be removed

On 2 December 2024 all teams will be deleted and the teams feature will be completely removed.

For more information about groups read the release announcement for groups, the teams documentation and Terraform migration guide.

The avn provision command has the new --google-privilege-bearing-service-account-id option, which is mutually exclusive with --aws-iam-role-arn, for example --google-privilege-bearing-service-account-id projects/your-project/serviceAccounts/cce-cce0123456789a@your-project.iam.gserviceaccount.com.

For more information, see Aiven CLI 4.2.1 release notes.

Discover Aiven for Valkey, the latest addition to our managed service portfolio. Aiven for Valkey is an open-source, in-memory NoSQL database designed as a seamless and reliable alternative to Redis OSS. Key features include:

• Open source initiative: Valkey is a fork of OSS Redis®, version 7.2.5, maintained by the Linux Foundation. It is fully backward compatible with Redis and licensed under the BSD-3 license.
• Redis compatibility: Fully compatible with Redis OSS 7.2.4 and Aiven for Caching.
• High performance: Fast data retrieval with low latency.
• Managed service: Aiven handles maintenance, updates, and monitoring.
• Migration support: Supports data migration from external Redis® and Valkey databases.

For more information, see Aiven for Valkey documentation.

This release also includes updated user config options, new ServiceIntegration fields, bug fixes, and other improvements.

You can view the full changelog on the Aiven Operator website.

h3-pg is a PostgreSQL extension that allows efficiently index and analyze geospatial data, including creating heatmaps, performing spatial analyses, and enhancing location-based decision-making for applications and services within the PostgreSQL database.

For more information, see the h3pg documentation.

• New
  • halfvec type
  • sparsevec type
  • Support for indexing the bit type
  • Support for indexing L1 distance with HNSW
  • binary_quantize function
  • hamming_distance function
  • jaccard_distance function
  • l2_normalize function
  • subvector function
  • Concatenate operator for vectors
  • CPU dispatching for distance functions on Linux x86-64
• Improved
  • Performance of on-disk HNSW index builds
  • Comparison operators to support vectors with different dimensions
• Fixed
  • Initialization fork for indexes on unlogged tables
  • undefined symbol error with GCC 8
  • Compilation error with universal binaries on Mac
  • Compilation warning with Clang < 14

This section is powered by Aiven AI Database Optimizer and it tracks slow queries for you and also suggests how to optimize them.

• Protobuf rendering: Fixed the boolean lowercase rendering in the java_multiple_files option for Protocol Buffers.

For a complete list of fixes and detailed information, see the Karapace release notes.

• Added support for controlling retry behavior in API calls.
• Corrected the help output for avn service list command.

For more information, see Aiven CLI release notes.

• Improve search and analytics performance with hybrid search functionality combining neural and lexical search, offering up to 4x performance increases.
• Gain real-time insights into query performance.
• Reduce memory footprint for FAISS engine indexes.
• Query and manage external data sources more efficiently with expanded functionality for multiple data sources.
• Support resiliency with I/O-based admission control.
• Boost performance for hybrid search use cases.
• Accelerate date histogram queries with multi-range traversal, especially benefiting time-series data analysis.
• Improve search efficiency with tiered caching within the request cache (Experimental).
• Access multiple data sources with consistent navigation and discovery tools.
• Filter k-NN results based on maximum distance or vector score.
• Build search pipelines with index alias support.

For more details on all the new features and improvements in OpenSearch 2.14.0, see the OpenSearch Release Notes 2.14.0. Apply the service maintenance updates to upgrade to this version. For more information, see Aiven for OpenSearch documentation.

• Avro/JSONSchema error handling: Fixed the issue where a payload that does not match the schema now returns a 422 error code, providing more precise error feedback.
• Schema message consumption: Enhanced schema message consumption to process multiple messages during startup, reducing overhead and increasing efficiency. After startup, it now consumes one record at a time to ensure consistent performance.
• Union type handling: Resolved an issue to correctly handle Avro array and map names in union types, ensuring proper data consistency.

For a complete list of fixes and detailed information, see the Karapace release notes.

What's changed:

• Name update: All references to Aiven for Redis® have been updated to Aiven for Caching.
• Licensing compliance: The service remains compatible with legacy Redis® OSS, and we will continue to provide updates for it until its end-of-life (EOL).

What's not changing:

• Service functionality: Existing services, API endpoints, CLI commands, and other functionalities remain unchanged. No immediate action is required from current customers.

For more information, see the Aiven for Caching documentation.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

To see the list of projects go to Projects > View all projects.

• New
  • Support for parallel index builds for HNSW
  • Validation for GUC parameters
  • Docker tags for each supported version of Postgres
  • Check for MVCC-compliant snapshot for index scans
• Improved
  • Fixed error with ANALYZE and vectors with different dimensions
  • Fixed segmentation fault with shared_preload_libraries
  • Fixed vector subtraction being marked as commutative
  • Reduced lock contention with parallel HNSW index builds
  • Storage for vector changed from extended to external
  • Improved performance of HNSW
  • Improved performance of HNSW index builds
  • Reduced memory usage for HNSW index builds
  • Reduced WAL generation for HNSW index builds
  • Docker image moved to the pgvector org
  • Dropped support for Postgres 11

Recommended actions:

• For Apache Kafka producer clients earlier than version 3.6.0:
  • Temporarily set transaction.partition.verification.enable to false in the service's advanced configuration during the upgrade.
  • Revert this setting to true after upgrading to Apache Kafka 3.6.0 or later.
• For Apache Kafka producer clients using version 3.6.0 or later:
  • Disabling transaction verification might lead to transaction duplicates.
  • Ensure transaction.partition.verification.enable is set to true after upgrading to version 3.6.0 or later.

For assistance or more details, contact the Aiven support team.

• Allows creating connectors in the STOPPED state for Apache Kafka® Connect (KAFKA-15470): Newly created connectors are no longer in the RUNNING state by default.
• Introduces cluster-wide dynamic log adjustments for Apache Kafka® Connect (KAFKA-15428): Now, with a single REST API request you can modify the logging level for the worker that receives the request or the logging level for all workers in the cluster.
• Implements client exponential backoff (KAFKA-9800) - the maximum amount of time in milliseconds to wait when retrying a request to the broker that has repeatedly failed. If provided, the backoff per client increases exponentially up to this maximum for each failed request.

To learn more about Aiven for Apache Kafka, see Aiven documentation.

Application users are a special type of user that super admin can create to use for programmatic access to Aiven through Terraform. They make it easier to audit and manage access to your Aiven resources for applications.

This release also includes fixes for the aiven_kafka_connector 201 error and the invalid token error.

For more information on Aiven Provider for Terraform, see the documentation.

Key updates:

• The connector is now compatible with Elasticsearch versions 7.x onwards.
• Elasticsearch 8.x no longer supports the _type parameter, which affects data structure and queries. Update your Elasticsearch queries and data handling accordingly. Compatibility with Elasticsearch versions 6.x and earlier has been discontinued due to deprecated features and compatibility issues.
• This update applies only to the Apache Kafka Connect connector and does not affect the metric agent.

Action required:

If you currently use Elasticsearch version 6.x or earlier, you must upgrade to at least Elasticsearch version 7.x by applying the service maintenance updates to continue using this connector.

For details on how to use this connector, see Create a sink connector from Apache Kafka® to Elasticsearch.

What’s new:

• Storage Write API support (beta): The Storage Write API is now available as an opt-in feature. To enable it, you can use the useStorageWriteApi property. By enabling this feature, the default stream will be used. However, if you set the enableBatchMode property to true, the pending type stream will be used instead. This is currently a beta feature, and its use in production environments is not recommended.

Bug fixes:

• Multiple topics routing: The topic2TableMap property now allows multiple topics to be routed to the same table.
• Retry logic for merge queries: Added retry logic for jobInternalError errors when performing merge queries in upsert/delete mode.

For more information, refer to the GitHub release notes for the BigQuery connector for Apache Kafka.

These users are centrally managed by organization super admin, who now have full visibility into how their organization’s resources are programmatically accessed on the Aiven platform. The application user tokens also add another level of security. Super admin can restrict users from creating personal tokens, ensuring that only approved applications access the Aiven API with application user tokens. This also makes it easier for super admin to regularly rotate tokens according to your organization’s security policies.

Find out more about using application users to securely manage access to the Aiven platform in the documentation.

Refer to the TimescaleDB release notes for 2.14.0, 2.14.1, and 2.14.2 to check all the changes introduced in these releases, including deprecation announcements and operational instructions.

Key new features

• Ability to change compression settings on existing compressed hypertables at any time: New compression settings take effect on any new chunks that are compressed after the change.
• Reduced locking requirements during chunk recompression
• Limiting tuple decompression during DML operations to avoid decompressing a lot of tuples and causing storage issues (100k limit, configurable)
• Helper functions for determining compression settings
• Plan-time chunk exclusion for real-time Continuous Aggregate by constify the cagg_watermark function call

More new features

• Add views for per chunk compression settings #6630
• Add plan-time chunk exclusion for real-time CAggs #6325
• Remove support for creating Continuous Aggregates with old format #6360
• Add functions for determining compression defaults #6386
• Remove multinode public API #6410
• Allow SQLValueFunction pushdown into compressed scan #6440
• Support approximate hypertable size #6463
• Make compression settings per chunk #6513
• Remove reindex_relation from recompression #6529
• Fix if_not_exists behavior for CAgg policy with NULL offsets #6531
• Remove restrictions for changing compression settings #6545
• Limit tuple decompression during DML operations #6566
• Change compress_chunk and decompress_chunk to idempotent version by default #6579
• Add LWLock for OSM usage in loader #6608
• Deprecate recompress_chunk #6609
• Add optional recompress argument to compress_chunk #6609

Bug fixes

• Fix segfault in cagg_validate_query #6655
• Fix refresh on empty CAgg with variable bucket #6660
• Don't try to compress osm chunks #6670
• Fix extension update of compressed hypertables with dropped columns #6636
• Reset sequence numbers on non-rollup compression #6637
• Disable default indexscan for compression #6639
• Fix DecompressChunk path generation with per chunk settings #6651
• Inefficient join plans on compressed hypertables #6541
• Enable now() plantime constification with BETWEEN #6491
• Fix create_hypertable referenced by fk succeeds #6494
• Suboptimal query plans when using time_bucket with query parameters #6498
• time_bucket_gapfill with timezones doesn't handle daylight savings #6507
• Make extension state available through function #6509
• Log extension state changes #6512
• Disallow triggers on CAggs #6522
• Reduce locking level on compressed chunk index during segmentwise recompression #6523
• Fix if_not_exists behavior for CAgg policy with NULL offsets #6531
• Fix pathtarget adjustment for MergeAppend paths in aggregation pushdown code #6571
• Fix compressed chunk not found during upserts #6575
• Fix recompression policy ignoring partially compressed chunks #6592
• Ensure qsort comparison function is transitive #6610

• Added the service integrations external_aws_cloudwatch_logs, external_elasticsearch_logs, external_opensearch_logs, and prometheus.
• Added a safety check for purging the IP filter list using the AIVEN_ALLOW_IP_FILTER_PURGE environment variable to restrict access to a service from the internet.
• Converted ip_filter_object into a set. This was one of the most requested features.

Additionally, with this release, the aiven_organization_user_group_member and aiven_organization_group_project resources are no longer in beta.

For more information on Aiven Provider for Terraform, see the documentation.

Key features and benefits:

• Redis compatibility at scale: Aiven for Dragonfly is a seamless drop-in replacement for Redis, capable of handling extensive workloads with enhanced performance.
• Optimized for large-scale operations: Dragonfly is specifically built to address the scalability and resource utilization limitations of Redis Open Source Software (Redis OSS).
• Seamless migration and compatibility: Mirgrate from Aiven for Redis to Dragonfly with no code modifications needed, maintaining the Redis experience while enhancing performance and scalability.
• Advanced performance and scalability: Dragonfly's unique threading and shared-nothing architecture enables it to scale vertically and handle over 1 TB workloads, with 10x throughput of Redis OSS and reduced latency.
• Reliability and support: Aiven for Dragonfly is fully supported by Aiven's service level agreements (SLAs), ensuring it can manage production workloads effectively.

To learn more, see Aiven for Dragonfly.

Key changes:

• For new setups: Debezium connectors on new Apache Kafka Connect services will default to version 2.5.
• For existing setups: Setups currently using Debezium version 1.9 will not be automatically upgraded. These setups will continue to use Debezium version 1.9, even during service maintenance.
• Configuration changes: Starting with version 2.0, Debezium introduced significant changes to configuration parameters. It is recommended to review and test these updates in a non-production environment. For detailed information, see the new configuration namespaces section.
• Plugin deprecation: With the introduction of Debezium version 2.5, the `wal2json` plugin is now deprecated.

Actions required:

• For new setups: No action is needed; Debezium version 2.5 will be used by default.
• For existing setups opting to upgrade:
  • A maintenance update is required for existing setups to upgrade to Debezium version 2.5. Without this maintenance, your setup will remain on version 1.9.
  • Test your configurations with version 2.5 in a non-production environment to evaluate compatibility.
  • Due to breaking changes, the connector may need to be reconfigured. To upgrade, submit a support ticket. It may be necessary to reconfigure the connector with the new settings.
• Plugin migration: Debezium version 2.5 has deprecated the wal2json plugin, and we advise users to transition to either the pgoutput or decoderbufs plugins as per Debezium's recommendations.

Support: If you need help with the upgrade process or have questions regarding Debezium version 2.5, submit a support ticket via our support channel.

Check all supported GCP cloud regions in List of available cloud regions.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

The new platform will allow us to rework the content more efficiently and to bring a more interactive experience to our users.

Bug fixes

• Use numrows_pre_compression in approximate row count
• Use processed group clauses in PG16
• Change bgw_log_level to use PGC_SUSET
• Disable vectorized sum for expressions
• Read CAgg watermark from materialized data
• Fix groupby pathkeys for gapfill in PG16
• Fix index matching during DML decompression
• Fix compressed chunk permission handling on PG16
• Fix lost concurrent CAgg updates
• Fix unique expression indexes on compressed chunks
• Fix use of freed path in decompression sort logic

With verified domains, existing organization users seamlessly transition into managed users. New users can be provisioned and managed through your identity provider.

Key new features

• Paxos v2
• Guardrails
• Client-side password hashing
• Partition denylist

This release comes with a wealth of new capabilities and performance enhancements:

• Better query planning and bulk loading performance
• New logical replication features
• Developer experience features
• Strengthened security and access control

For more information on what PostgreSQL version 16 introduces, check PostgreSQL 16 release notes.

Other enhancements include:

• Create general tickets that aren’t linked to a specific project or service
• Add files like screenshots and logs directly to the tickets
• View and follow all tickets in your organization
• Escalate issues to your account team

Check out the documentation for details on how to use these and other support features.

• Enhanced capabilities to manipulate and refine search results are now generally available with the release of search pipelines.
• The Neural Search Plugin, integrating machine learning language models to improve search accuracy and relevance, is now generally available.
• The Machine Learning Framework is now generally available in OpenSearch Service, offering full integration and simplifying the incorporation of external models for data analysis.
• Enhanced vector database functionality upgraded with Facebook AI Similarity Search (FAISS) for more efficient query pre-filtering, improving search precision.
• Lucene byte-sized vectors now have advanced support, improving data storage and retrieval efficiency.
• Improved index compression codec for efficient data handling and enhanced performance.
• Support for geo_bounds, geo_hash, and geo_tile aggregations added to enhance geospatial data analysis.
• The new and improved hybrid query system and a built-in integrated normalization processor help enhance the relevance of search results.
• Sparse retrieval for text-based vector search is now available, offering more focused and efficient search capabilities.
• The OpenSearch search comparison tool, now widely available, allows for comprehensive and detailed analysis of search results.
• OpenSearch Dashboards enhancements:
  • Removal of AngularJS dependencies for platform modernization.
  • Introduction of a new visual theme.
  • Enhanced availability during minor version upgrades.

For a comprehensive understanding of all the new features and improvements in OpenSearch 2.11.1, refer to the OpenSearch Release Notes 2.11.1. To upgrade to this version, apply the service maintenance updates. For information on Aiven for OpenSearch, see the documentation.

Key new features

• Full PostgreSQL 16 support for all existing features
• Vectorized aggregation execution for sum()
• Track chunk creation time used in retention / compression policies

More new features

• Add chunk-wise sorted paths for compressed chunks
• Simplify hypertable DDL API
• Reduce WAL activity by freezing compressed tuples immediately
• Add metadata for chunk creation time
• Make Continuous Aggregates materialized only (non-realtime) by default
• Change show_chunks/drop_chunks using chunk creation time
• Show batches / tuples decompressed during DML operations in EXPLAIN output
• Keep track of catalog version
• Use creation time in retention/compression policy
• Add SQL function cagg_validate_query

Bug fixes

• Add GUC for setting background worker log level
• Allow enabling compression on hypertable with unique expression index
• Check if worker registration succeeded
• Fix exception detail passing in compression_policy_execute
• Fix missing bms_del_member result assignment
• Fix negative bitmapset member not allowed in compression
• Potential data loss when compressing a table with a partial index that matches compression order
• Add support for startup chunk exclusion with aggs
• Repair relacl on upgrade
• Fix segfault when creating a cagg using a NULL width in time bucket function
• Make timescaledb_functions.makeaclitem strict
• Fix typmod and collation for segmentby columns
• Fix tablespace with constraints
• Enable segmentwise recompression in compression policy

Key new features

• Updated list of table functions and engines supported in Aiven for ClickHouse®
• New capabilities for data ingestion, such as input_format_csv_allow_variable_number_of_columns or s3_skip_empty_files and engine_url_skip_empty_files for skipping empty files in s3 and url table functions
• Improvements for working with metadata and .parquet files in particular, such as faster reads and count operations

Bug fixes

• Fixes improving security
• Fixes improving performance

Check all supported GCP cloud regions in List of available cloud regions.

Service contacts receive all notifications for that service. If no service contacts are set, email notifications will be sent to the project contacts. The default project contacts are the project admins and operators.

Using these settings, you can ensure the right people receive important updates and avoid overwhelming others with unnecessary emails.

Learn more about notifications and how to set contacts in the documentation.

For an added layer of security, you can also enforce two-factor authentication for password logins.

Read more about the authentication types and learn how to set an authentication policy in the documentation.

To ensure you are using supported versions with the latest functionality, the EOL dates are published in the documentation.

• Upgraded to Aiven Commons library version 0.12, featuring a new topic-partition-key record grouper for improved record organization.

For detailed information, refer to the GitHub release notes and documentation for Aiven’s GCS Sink Connector for Apache Kafka.

Bug fixes

• Align gapfill bucket generation with time_bucket
• Ensure fixed_schedule field is populated
• Fix EXPLAIN ANALYZE for compressed DML

For detailed instructions, see Enable OAuth2/OIDC support for Apache Kafka® REST proxy.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

This version includes breaking changes and soon all previous versions will return incorrect content. In addition, updates continue to be made to support organizations, including:

• The avn card commands were deprecated in version 3.0.0 and have now been removed.
• The avn account team commands are deprecated and will be removed in the next version.
• The --parent-id is now mandatory in the avn project create command.
• The following have been removed from avn project create/update:
  • --account-id
  • --billing-address
  • --billing-currency
  • --billing-email
  • --billing-extra-text
  • --card-id
  • --country-code
  • --vat-id
• The detailed billing related information is replaced by billing group details in the avn project list command.

New feature:

• Added a validator for S3 bucket names to ensure compliance with AWS's specific naming standards.

Improvements and fixes:

• Upgraded to Aiven Commons library version 0.12, featuring a new topic-partition-key record grouper for improved record organization.
• Improved the documentation for grouping records and made it more user-friendly, providing clear guidance to assist users.
• Corrected log messages to accurately reflect the connector's start process.

For more details, see the S3 connector for Apache Kafka changelog and the S3 connector for Apache Kafka documentation.

Key highlight: Kafka tiered storage

With the introduction of KIP-405, Apache Kafka now supports tiered storage. This feature allows for segregating storage into different tiers based on the data access pattern, optimizing storage costs and performance. This is a limited availability feature and is recommended for non-production environments. To learn more, see Tiered storage in Aiven for Apache Kafka®.

Other features & improvements:

• Offset reset capability in Kafka Connect (KIP-875): Kafka Connect now supports offset reset capability, ensuring more reliable data streaming and handling.
• Rack aware task assignment in Kafka Streams (KIP-925): Kafka Streams has introduced a rack-aware task assignment feature, enhancing stream task distribution and fault tolerance. This feature is partially implemented as of this release.
• Transactions server-side defense (Part 1)(KIP-890): Addresses hanging transactions impacting read committed consumers and preventing compacted logs from being compacted.
• Accept duplicate listener on port for IPv4/IPv6 (KIP-797): Brokers can now be configured with listeners with the same port on different IP stacks, simplifying network configurations.
• Grace Period in Stream Table Join (KIP-923): Kafka Streams now offers a grace period for stream-table joins, providing more flexibility in handling late-arriving records.
• Modernize Connect plugin discovery (KIP-898): Connect workers can now access ServiceLoader manifests and module information during startup, streamlining the plugin class discovery process. This update also allows connector developers to incorporate service declarations into their plugins.
• Various bug fixes and improvements have been made to enhance the stability and performance of Apache Kafka.

To learn more about Aiven for Apache Kafka, see Aiven documentation.

Features

• Insert into index during chunk compression
• MERGE support on hypertables
• Make hypertables support replica identity
• Index scan support during UPDATE/DELETE on compressed hypertables
• Support for partial aggregations at chunk level
• Enable ChunkAppend for partially compressed chunks
• Improve the number of parallel workers for decompression
• Enable altering job schedule type through alter_job
• Make logrepl markers for (partial) decompressions
• Relax invalidation threshold table-level lock to row-level when refreshing a Continuous Aggregate
• Support CAgg names in chunk_detailed_size
• Make set_chunk_time_interval CAggs aware
• Allow ALTER TABLE ... REPLICA IDENTITY (FULL|INDEX) on materialized hypertables (continuous aggregates)
• Add job exit status and runtime to log
• CREATE INDEX ONLY ON hypertable creates index on chunks

Bug fixes

• Fix planner distributed table count
• Avoid decompressing batches using an empty slot
• Fix concurrency errors in OSM API
• Do not throw an error when deprecation GUC cannot be read
• Fix interval calculation for hierarchical CAggs
• Check unique indexes when enabling compression
• _timescaledb_internal.create_compressed_chunk doesn't account for existing uncompressed rows
• Move functions to _timescaledb_functions schema
• Chunk_create must add an existing table or fail
• Fix duplicates on partially compressed chunk reads
• Fix crash in COPY from program returning error
• Place data in first/last function in correct mctx
• Call eq_func correctly in time_bucket_gapfill
• Correct row count in EXPLAIN ANALYZE INSERT .. ON CONFLICT output
• Fix server crash on UPDATE of compressed chunk
• Fix server crash when using duplicate segmentby column
• Fix segfault in set_integer_now_func
• Fix approximate_row_count for CAggs
• Improve compressed DML datatype handling
• Propagate parameter changes to decompress child nodes
• Schedule compression policy more often

New features

Introduces support for protocol-level named prepared statements: Using prepared statements together with PgBouncer can reduce the CPU load on your system.

Changes

• Improves security for default TLS settings and changes required for the FIPS compliance.
• Discontinues support for UDNS.

Bug fixes

• Fixes the crash with error message FATAL in function client_proto(): bad client state: 6/7 [#928] (the bug introduced in 1.18.0).
• Fixes the crash with error message FATAL in function server_proto(): server in bad state: 11 [#927] (the bug introduced in 1.18.0).
• Reduces cancellation sending log level [#903].
• Fixes the slog log prefix for peers [#922].
• Fixes typos in the docs [#932].
• Fixes errors pointed out by the static analyzer [#943].
• Prevents killing all waiting clients on temporary FATAL errors during logging in [#946].
• Enables using the auto-database when a database in auth_dbname is not explicitly configured [#921].

For more information, see OpenSearch® Dashboards.

• Canvas panel: A dynamic platform for data visualization is now generally available.
• Enhanced data source selection: A revamped process for data source selection, making it more intuitive and efficient.
• Time series regional support: Introducing time regions to the time series panel for improved data representation.
• Annotation filtering: Refine and filter dashboard annotations for a clutter-free view.
• Revamped log context: Experience a redesigned log context interface optimized for consistency and user-friendliness.
• Multi-data source queries in Explore: Dive deeper by querying multiple data sources concurrently in Explore.

For a complete list of new features and improvements, see Grafana v10.0 documentation.

The following fields in the aiven_project resource have been deprecated in this release:

• add_account_owners_admin_access
• use_source_project_billing_group

This release also includes several fixes. For details, view the changelog.

Key features

• Feature flags for TimescaleDB features

Bug fixes

• Fix DISTINCT query with JOIN on multiple segmentby columns
• Fixed two bugs in decompression sorted merge code
• Ensure pg_config --cppflags are passed
• Fix quoting owners in sql scripts
• Fix crash in 1-step integer policy creation

Organizational units

Organizational units give you another level of hierarchy to organize your projects and services. These can be used, for example, to easily split production and testing environments.

Learn more about how you can use organizations and organizational units to create a hierarchical structure that fits your needs.

Organization users and groups

To reduce the administrative burden of managing users across many projects, users are now centrally managed at the organization level and can be added to groups. Organization admin invite users directly to an organization and then group them together, making it easier to control user access to projects and services.

Get started with organization users by inviting users to your organization and creating groups.

• Data consistency issue fixed (SNOW-811265): This version fixes a problem where revoking specific partitions would clear the entire partitionToChannel map, increasing the risk of data inconsistencies. Now, only the partitions that have been revoked are removed.
• New cache for schema queries (SNOW-858328): A cache has been added for schema evolution permission queries, reducing the number of queries sent to Snowflake and improving the connector's throughput.
• Metrics enhancement: Minor adjustments to Snowpipe telemetry and JMX metrics have been made, providing more precise and useful data regarding the connector's performance and operational status.
• Log level change: The Pipe already registered message has been updated from an error to a warning level, indicating it is a non-critical issue that does not require any configuration changes.

For a complete list of changes, see the changelog on GitHub.

Key features

• New
  • HNSW index type
  • Support for parallel index builds for IVFFlat
  • l1_distance function
  • Element-wise multiplication for vectors
  • sum aggregate
• Improved
  • Performance of distance functions
  • Error message for malformed vector literal
  • Cost estimation
  • Support for spaces with text input

Bug fixes

• Fixed out of range results for cosine distance
• Fixed results for NULL and NaN distances for IVFFlat
• Fixed segmentation fault with index creation on i386
• Fixed segmentation fault with text input
• Fixed consecutive delimiters with text input
• Fixed infinite and NaN values with binary input
• Fixed infinite values with vector addition and subtraction
• Fixed infinite values with list centers
• Fixed compilation error when float8 is pass by reference
• Fixed compilation error on PowerPC

Key features:

• Role-based access controls: Configure advanced security measures, including document-level and field-level security, along with user and role mapping to safeguard sensitive data.
• SAML integration: Gain seamless access to your OpenSearch Dashboard via your preferred identity provider with our basic SAML integration.
• OpenID Connect integration: Enhance your authentication options by securely authenticating and authorizing users through OpenID Connect.
• Multi-tenancy support: Utilize OpenSearch Dashboard multi-tenancy to create isolated spaces for different tenants, each with unique permissions and data access levels.
• Audit logging: Maintain a comprehensive audit trail with OpenSearch audit logs for compliance and policy enforcement.

For an in-depth understanding of OpenSearch Security, refer to OpenSearch Security for Aiven for OpenSearch® documentation.

The following properties have also been added in this version:

• scheduled_rebalance_max_delay_ms property for aiven_kafka and aiven_kafka_connect
• openid property for aiven_opensearch
• auto_login property for aiven_grafana

Additionally, this release includes fixes and improvements, including an enhanced go client with advanced retries for better stability.

For more information on Aiven Provider for Terraform, check out the documentation.

• Advanced retries in the go client for better stability
• Support for Kafka 3.5

For details on the updates and deprecations, check out the complete Aiven Operator changelog.

New navigation

Effortlessly navigate your projects and services via the restructured left-side panels.

Breadcrumbs

Jump directly to what you need in your project structure using the new breadcrumbs navigation.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

Check out Apache Cassandra release notes for more details about the Apache Cassandra minor releases.

Key features:

• Improved stability with KIP-903: This feature prevents replicas with stale broker epochs from joining the ISR, thereby improving the overall stability of the clusters.
• Full support for distributed mode in MirrorMaker 2.0 clusters (KIP-710): MirrorMaker 2.0 now fully supports dynamic topic and group filter changes in distributed mode, resolving a previous issue where MirrorMaker would not react properly to these changes.
• First-class offset support in Kafka Connect (KIP-875): Kafka Connect now supports resetting offsets, allowing logs to be reproduced from the beginning if something goes wrong.
• Enhancements to Kafka Streams: Kafka Streams now have better exception handling with KIP-399 and a versioned state store with KIP-889.

Bug fixes:

• Fix for CooperativeStickyAssignor (KAFKA-14639): This fix addresses an issue with the CooperativeStickyAssignor that could lead to unexpected results when running multiple consumers in parallel.
• Fix for rebalancing issues (KAFKA-13891): This fix addresses a bug that could cause consumers to rebalance many rounds before becoming stable.
• Fix for MirrorMaker 2.0 offset sync issue (KAFKA-12558): This fix addresses a race condition in MirrorSourceTask where certain partition offsets may never be sent.

To learn more about Aiven for Apache Kafka, see Aiven documentation.

• Organizations
  • avn organization create
  • avn organization list
  • avn organization update
  • avn organization delete
• Organization users
  • avn organization user invite
  • avn organization user list
• Payment cards
  • avn organization card create
  • avn organization card list
  • avn organization card delete

The avn card commands are deprecated and will be removed in the next major version.

There are also important changes to the avn project create command:

• Added --parent-id, which takes the organization ID or account ID. This will be mandatory in the next major version.
• Deprecated the billing information parameters since these are now used with avn billing-group. These will be removed in the next major version:
  • --country-code
  • --billing-address
  • --billing-extra-text
  • --billing-currency
  • --vat-id VAT_ID
  • --billing-email

Check out our documentation to learn more about organizations.

For more information and an example of how to use this resource, view the documentation.

Key highlights:

• You can now send DebeziumTimeStamp fields to BigQuery as they are without any conversion.
• You can use the KIP-610 feature to route errant records to a Dead Letter Queue (DLQ) for later inspection and processing.
• Added a retry mechanism to handle concurrent serialization errors that may occur when merging data in upsert/delete mode.

For a complete list of changes, see the Confluent changelog. To upgrade to this version, apply the service maintenance updates.

New navigation

Effortlessly navigate your projects and services via the restructured left-side panels.

Breadcrumbs

Jump directly to what you need in your project structure using the new breadcrumbs navigation.

Now, you can activate and try out the new navigation before it's officially launched. You can switch back to the original navigation in your profile settings > Feature preview.

A new resource, aiven_organization_user has also been added for managing users at the organization level.

• Scheduler accidentally getting killed when calling delete_job
• Fix Result node handling with ConstraintAwareAppend on compressed chunks
• Ensure tlist is present in decompress chunk plan
• Fixed handling of NULL values in bookend_sfunc
• Fixed batch look ahead in compressed sorted merge
• Mark cagg_watermark function as PARALLEL RESTRICTED
• Copy job config JSONB structure into current MemoryContext
• Improve continuous aggregate query chunk exclusion

• Consumer Fetch Response: Added a timestamp to the consumer fetch response to provide more detailed information about data retrieval.
• Exception Handler: Fixed an issue with the exception handler incorrectly fetching raw schema after deletion.
• Partitioning: Improved the partitioner to now accurately assign partitions randomly when a key isn't provided.
• Inconsistent Schema: Fixed an issue with inconsistent schema during message production.

For a complete list of fixes and detailed information, see Karapace release notes.

• New Single Message Transform (SMT) implementation: Introducing ExtractTopicFromValueSchema, a new SMT allowing more flexible topic management by extracting topics from value schemas.
• Extended support for schema-less values in ExtractTopic transform: The ExtractTopic transform now provides extended support for schema-less values, including maps and primitive values, in addition to schema-based values
• Dependency Upgrades: This release includes multiple dependency upgrades.

For detailed information, refer to GitHub Release.

New Feature:

• Introduced a new master coordinator that uses the `aiokafka` library, enhancing performance and reliability by improving server change handling and enabling asynchronous operations.

Enhancements and fixes:

• Fixed cache management to selectively update only required metadata segments instead of full replacement.
• Fixed offset management to move offsets consistently, even with invalid JSON records, preventing processing delays.
• Refactored schema version resolution and validation logic, including support for '-1' as the latest version.
• Refactored version class to remove integer inheritance, enhancing its structure for stronger typing and robustness.
• Increased test coverage to encompass JSON schema validation across varied subjects.
• Implemented tests for storing soft-deleted schemas post-compaction to ensure accessibility and resolve previous visibility issues.
• Removed `kafka-python` dependency, transitioning error mappings to the `aiokafka` library.

For a complete list of fixes and detailed information, see the Karapace release notes.

• Query your data across OpenSearch clusters with PPL: PPL (Piped Processing Language) is a new query language that allows you to run SQL-like queries on your OpenSearch data. You can also use PPL to query data from multiple OpenSearch clusters, making it easier to analyze and visualize data from different sources.
• Simplify index management tasks: You can now use index templates and component templates to define settings, mappings, and aliases for your indices.
• Search pipelines experimental feature: Search pipelines let you transform and enrich your search results with custom logic. You can use search pipelines to modify the relevance score, add or remove fields, or apply any other logic that suits your use case.
• Enhancements to existing features: Existing features have been enhanced, including support for the latest version of OpenSearch Dashboards, default activation of cross-cluster replication, and an upgrade to the underlying OpenSearch version.

To upgrade to this version, apply the service maintenance updates. For information on Aiven for OpenSearch, see the documentation.
To learn more, see OpenSearch and OpenSearch Dashboards 2.8.0 Release Notes.

pgvector is a PostgreSQL extension that allows efficient handling of high-dimensional vector data within the PostgreSQL database for tasks such as similarity search and machine learning.

For more information, see the pgvector documentation.

Key features

• Support for DML operations on compressed chunks
  • UPDATE/DELETE support
  • Support for unique constraints on compressed chunks
  • Support for ON CONFLICT DO UPDATE
  • Support for ON CONFLICT DO NOTHING
• Join support for hierarchical Continuous Aggregates

All features

• Allow pushdown of reference table joins
• Improve Realtime Continuous Aggregate performance
• Improve unique constraint support on compressed hypertables
• Support UPDATE/DELETE on compressed hypertables
• Enable JOINS for Hierarchical Continuous Aggregates
• Add parallel support for partialize_agg()
• Refactor and optimize distributed COPY
• Add support for ON CONFLICT DO UPDATE for compressed hypertables
• Skip Ordered Append when only 1 child node is present
• Propagate vacuum/analyze to compressed chunks
• Reduce decompression during constraint checking
• Optimize compressed chunk resorting
• Support sending telemetry event reports

Bug fixes

• Fix SEGMENTBY columns predicates to be pushed down
• Handle user-defined FDW options properly
• Decompression may have lost DEFAULT values
• Fix issue creating dimensional constraints
• Improve interpolate error message on datatype mismatch
• Fix unique constraint on compressed tables
• Add permission checks to run_job()
• Enable run_job() for telemetry job
• Fix on-insert decompression after schema changes
• Quote username identifier appropriately
• Fix tablespace for compressed hypertable and corresponding toast
• Fix ALTER TABLE SET with normal tables
• Reduce memory usage for distributed analyze
• Fix subtransaction resource owner
• Fix DISTINCT query with JOIN on multiple segmentby columns

Important: This connector version includes breaking changes on the code level. If you have manually created data stream integrations using the previous version, these changes may impact your existing integrations.

Key highlights:

• Dead-letter Queue support: This release introduces support for the Dead-letter queue, which captures events that failed to be indexed in OpenSearch. The Dead-letter queue enhances fault tolerance and error handling by providing a separate queue for troubleshooting and analyzing indexing failures. For more information, see Error Handling in Connect.
• Configuration settings updated: The configuration settings related to handling rejected records and version conflicts in OpenSearch have been updated.

For more information, see release notes and OpenSearch Sink Connector Configuration Options.

More information on billing is available in the documentation.

New features

• Support for n:1 topics:table (SNOW-514303): The connector now supports mapping multiple Kafka topics to a single Snowflake table.
• Schema mapping (SNOW-625820): The connector has enhanced support for schema mapping, allowing you to map Kafka topic schemas to Snowflake table schemas, providing flexibility and control over how your data is structured in Snowflake.
• Auto table creation (SNOW-628677): This new feature enables the automatic creation of Snowflake tables to store Kafka data, reducing manual configuration requirements.

Configuration changes

• Exactly once for streaming ingest (SNOW-536429): Configuration support has been introduced for exactly-once semantics during streaming data ingestion, guaranteeing data consistency and integrity.

Improvements in error handling

• Offset token retries (SNOW-543664 and SNOW-451197): The handling of failed offset tokens has been improved, including incorporating automatic retries and offset replays for heightened reliability.
• Handling of failed insert rows and delivery to Dead-Letter Queue (DLQ) (SNOW-530369): Improved failure handling for row insertion has been implemented along with the new functionality to deliver failed records to a DLQ.
• Implementation of ErrantRecordReporter (SNOW-530369): Kafka records from the sink connector can be transferred to the DLQ for further analysis and reprocessing.

In addition to the above, there have also been updates to dependencies and multiple bug fixes to improve the overall performance and stability of the Snowflake Kafka Connector. For a comprehensive overview of all the changes, see the changelog on GitHub. To learn how to use this connector, see Configure Snowflake for a sink connector.

In your user profile, the new organizations tab has a list of all of your organizations. The create organization button has also moved from the organizations drop-down menu to the user profile.

Check out the documentation to learn more about organizations and get instructions on how to create organizations and organizational units.

The following are deprecated:

• aiven_account_team_project resource
• aiven_account resource
• account_id Use id instead.
• primary_billing_id
• owner_team_id
• is_account_owner

• Updated OpenSearch reporting plugin engine: The OpenSearch reporting plugin engine has been updated to enhance the accuracy and efficiency of data reporting.
• CRUD operations on indices with OpenSearch Dashboards: You can now create, read, update, and delete (CRUD) operations on indices directly through OpenSearch Dashboards. This gives you more control over your data management in one convenient location.
• Jaeger data source for Dashboards observability feature: Jaeger has been introduced as a new data source in your OpenSearch Dashboards. You can now monitor and troubleshoot transactions in complex distributed systems using this powerful distributed tracing system.
• Enhancements to OpenSearch Dashboards maps: There have been substantial enhancements to the OpenSearch Dashboards maps feature. You can now benefit from improved geospatial data visualization, more layer types, and better performance.
• Bug fixes and general improvements: Addressed various bugs and made general improvements across OpenSearch, Dashboards, and plugins. These fixes and enhancements aim to provide a smoother and more reliable experience while using OpenSearch on the Aiven platform.

To learn more, see OpenSearch and OpenSearch Dashboards 2.7.0 Release Notes. For information on Aiven for OpenSearch, see the documentation.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

Bug fixes

• Fix parameterization in DecompressChunk path generation
• Fix broken CAgg with JOIN repair function
• Fix file trailer handling in the COPY fetcher
• Add checks for malloc failure in libpq calls
• Out of on_proc_exit slots on guc license change
• Use consistent snapshots when scanning metadata
• Do not segfault on large histogram() parameters
• Ensure superuser perms during copy/move chunk
• Fix when no FROM clause in continuous aggregate definition
• Fix join rte in CAggs with joins
• Fix duplicated entries on timescaledb_experimental.policies view
• Fix segfault after column drop on compressed table
• Copy scheduled_jobs list before sorting it
• Allow named time_bucket arguments in Cagg definition
• Fix refresh from beginning of Continuous Aggregate with variable time bucket
• Use regrole for job owner
• Enable indexscan on uncompressed part of partially compressed chunks

• Kerberos authentication support: The upgraded connector now offers support for Kerberos authentication, providing users with a more accessible and secure way to authorize their Kafka streams with Splunk.
• Protobuf support: This connector version includes Protobuf support, simplifying the process of handling and transferring messages in the Protobuf format within the Kafka system.
• Timestamp extraction capability: This upgraded connector introduces the capability to extract specific timestamps from Kafka messages, allowing for precise timestamp-based operations.

For more information, see Splunk Connect for Kafka release notes. For information on how to use this connector, see Create a sink connector from Apache Kafka® to Splunk.

• Security fixes: Includes multiple CVEs (CVE-2022-35977, CVE-2023-22458, CVE-2023-25155, CVE-2022-36021, CVE-2023-28425, and CVE-2023-28856) addressing vulnerabilities related to integer overflow, denial-of-service attacks, assertion failures, and unauthorized access. These fixes enhance the security and stability of Redis.
• Performance improvements: Several commands and operations have been optimized to enhance speed and efficiency, leading to faster execution and improved overall performance.
• Enhanced functionality: The latest release introduces a variety of new commands and features that enhance the system's capabilities. These include commands for list element manipulation, finding the longest common subsequence between strings, claiming pending stream messages, and managing ACL rules through files.
• Bug fixes: This release includes multiple bug fixes, resolving issues such as memory leaks, replication failures, cluster crashes, and client tracking errors.

For more information, see Redis releases notes.

You can now upgrade your existing Aiven for Apache Kafka service to version 3.4.0 using the Aiven console. For more information on the upgrade procedure, see Apache Kafka® upgrade procedure.

To learn more, see Create a PostgreSQL® CDC connector-based Apache Flink®.

Prior to this release, using some service secrets at the same time could cause collisions between secret names. For example, using both Redis and Kafka secrets for the same application caused collisions for the HOST and PORT variables. The same was true when using two different Redis services for one application.

Now you can define a custom prefix for the secrets' keys in the connInfoSecretTarget. ​By default, the prefix is the kind name in all caps followed by an underscore.

​For backwards compatibility, this release keeps all the old variable names, but these ​will be removed in a future release. Be sure to prepare your applications for this change.

For more information and example files, see the API Reference documentation.

• Unqualified table names in the source connector are now enabled, making it easier to configure and use. A new property, table.names.qualify, is introduced with a default value of true. When set to false, the connector no longer fails if it detects multiple tables with the same unqualified name, and queries are run against the database using unqualified table names.
• Various dependencies have been upgraded to their latest versions.

For more information, see release notes and documentation.

This version also includes new annotations and labels fields for the connInfoSecretTarget secret feature.

For complete details, check out the Aiven Operator changelog.

• Upgraded dependencies to their latest versions.
• Bug fixes related to the MQTT Connector, resulting in a more stable and reliable data transfer between MQTT and other systems.
• This release does not include any breaking changes.

For more information, see Stream Reactor release notes. For a list of Stream Reactor connectors supported in Aiven, see Source connectors and Sink connectors.

• Aiven API updates for both new and dropped deprecations
• New validations and constraints for an improved user experience

The documentation continues to be improved, too, with more examples added to the API Reference and a new recipe for creating a PostgreSQL read-only replica.

For details on the deprecations and fixes, check out the complete Aiven Operator changelog.

To address this vulnerability, we recommend upgrading your Aiven for Grafana service to version 9.3.11 by performing a maintenance update. For more information on the security fixes, see Grafana security release.

Key features

• Joins in continuous aggregates that are defined over hypertables (upcoming in a follow-up release: the support for joins in hierarchical continuous aggregates)
• Re-architecture of how compression works: ~2x improvement on the INSERT rate into compressed chunks.
• Full PostgreSQL 15 support for all existing features (upcoming in a follow-up release: the support for the newly-introduced MERGE command on hypertables)

Bug fixes

• Support continuous aggregates names in hypertable_(detailed_)size
• Fix concurrent locking with the chunk_data_node table
• Fix incorrect memory handling
• Use NameData and namestrcpy for names
• Set PortalContext when starting a job
• Fix the uninitialized bucket_info variable
• Make the copy fetcher more asynchronous
• Fix the num_chunks inconsistency in the hypertables view
• Fix the column names handling in old-style continuous aggregates
• Fix the multi-node DML HA performance regression
• Fix the hierarchical continuous aggregates chunk_interval_size
• Fix sub-second intervals in hierarchical continuous aggregates
• Fix the use of the prepared statement in the async module
• Add a role-level security to the job error log
• Fix the next_start calculation for fixed schedules
• Fix the enabling of compression on continuous aggregates with columns requiring quotation marks

For more information, see the Google Cloud Pub/Sub Group Kafka Connector GitHub project. For information on using this connector, see Create a Google Pub/Sub source connector to Apache Kafka® and Create a sink connector from Apache Kafka® to Google Pub/Sub.

Some of the changes introduced in version 1.9 are:

• Deprecated copy.existing property and related properties. Instead, use startup.mode=copy_existing and related properties.
• Added change.stream.full.document.before.change setting to access and configure the pre-image of an update operation in the change stream event document.
• Added publish.full.document.only.tombstones.on.delete setting to send tombstone events when documents are deleted. This setting only applies when publish.full.document.only is true.

For more information on the upgrade and a complete list of changes, see the official release notes.

For information on how to use this connector, see Create a source connector from MongoDB to Apache Kafka® and Create a sink connector from Apache Kafka® to MongoDB.

• Improved error handling when rejecting invalid input.
• Fixed malformed request handling in subject and compatibility endpoints.
• Fixed handling of invalid schema on REST proxy publish.
• Fixed cancellation handling due to client disconnect.

For a complete list of fixes and detailed information, see Karapace release notes.

Removing bugs and security vulnerabilities increases database performance, improves service availability, and helps ensure business continuity.

See PostgreSQL release notes for more details about the PostgreSQL® minor releases.

• Quick access to recent projects
• Project health insights with service-level details
• Platform status updates
• Product updates for the Aiven platform

To explore the Home view, select Home from the top-level navigation in Aiven Console. If you still don’t have an organization on Aiven, create one to be able to check out this new feature.

For all existing Aiven for Grafana services older than 9.3.8, the update to 9.3.8 will be applied as a mandatory maintenance update in the next maintenance window. For more information on the security fixes, see Grafana security release.

In addition, support for service integrations have been added for:

• ClickHouse and PostgreSQL
• ClickHouse and Kafka
• AWS CloudWatch

There are two notable changes for Kafka-specific configuration: min_cleanable_dirty_ratio and metadata.name.

You can now set the min_cleanable_dirty_ratio Kafka topic configuration parameter with Aiven Operator for Kubernetes. This lets you control how frequently the log compactor cleans the log.

The metadata.name field is being replaced by the new KafkaTopic.Spec.topicName field. The new field supports additional characters and longer names. It will be required in a future release.

The documentation has also been overhauled to provide a better experience. Check it out for more information on the new features.

• Support for data streams has been added to optimize time-series data management. This simplifies ingesting continuously generated data such as logs, events, and metrics.
• The connector now features index.write.method for improved OpenSearch index control. By default, the ``index.write.method`` is set to insert, which replaces any existing record with the same document ID. Alternatively, users can also use upsert to create a new document if one does not exist or update an existing one.

Important: This version includes breaking changes on the code level, which may impact existing data stream integrations that were created manually.

For more information, see release notes and OpenSearch sink connector configuration options.

• Improved backup creation and restoration: The backup restoration process has been updated to wait for the message production to complete and will now raise an error if the message is not successfully sent to the restored topic.
• Improved JSON serialization and deserialization handling: Improved handling for deserialization failures during record consumption, including better error messages and the ability to catch JSON decode errors.
• Avro enum compatibility: Fixed the compatibility handling for Avro enum symbol additions and removals and enhanced the compatibility result logging for better informative insights.
• Support for UltraJSON: Karapace now supports using UltraJSON for improved performance.

To view the complete list of fixes in this version, see Karapace release notes.

• Enhancements made to cluster config fields for OpenSearch Performance Analyzer
• Windows and Mac CI for OpenSearch Anomaly Detection have been included
• Updates include bug fixes, infrastructure updates, and maintenance enhancements

For detailed information on the updates in OpenSearch 1.3.8, see OpenSearch release notes and Aiven for OpenSearch documentation.

For more information on Aiven Provider for Terraform, see the documentation.

• schema fields use strict types instead of string
• support for strict types in diff functions

These deprecated resources have also been removed:

• aiven_database
• aiven_service_user
• aiven_vpc_peering_connection
• aiven_flink_table
• aiven_flink_job

For more information on Aiven Provider for Terraform, see the documentation.

Organizations and units also provide:

• Better management of your environments by separating production and testing projects
• Set-and-forget configuration for things like authentication methods and billing
• Simpler management of your projects by centralizing administration

For more information on organizations and units see the blog post or the documentation.

This release contains the following bug fixes on top of release 2.9.2:

• Skip bucketing when start or end for a refresh job is null
• Fix column ordering in the compressed table index not following the order of a multi-column segment by definition
• Don't enable clang-tidy by default
• Fix year not being considered as a multiple of day/month in hierarchical continuous aggregates
• Lock down search_path in SPI calls

Key highlights of features and enhancements in this release include:

• Easy integration: Connect to Aiven for Apache Kafka, Aiven for PostgreSQL, and Aiven for OpenSearch with just a few clicks in the Aiven Console.
• Aiven for Flink® applications: Offers a layer of abstraction that includes source and sink tables definitions, SQL transformation definitions, and deployment parameters.
• Interactive queries: Preview data of a Flink table or job with our built-in SQL editor without the need to output it to a sink table. Enhance your developer experience with the convenience of our interactive query feature.
• Deployment stage: Allows you to define execution parameters and reuse savepoints to start an application from an existing state.
• Track Changes with versioning: Create a new application version every time you change the table or data transformation. Quickly revert to a previous version if needed.
• Improved Checkpoint and Savepoint: Automates the backup of pipeline state and provides graceful shutdown and startup.
• Enhanced connectors: Improves the set of connection options, allowing you to stream the result of the data pipeline directly to an OpenSearch index or notify people in Slack.

For more information, see Aiven for Apache Flink documentation.

Key features introduced with this release are the following:

• Hierarchical Continuous Aggregates (aka Continuous Aggregate on top of another Continuous Aggregate)
• Improved time_bucket_gapfill function to allow specifying the timezone to bucket
• Fixed schedules for background jobs and the ability to check job errors
• alter_data_node() to change the data node configuration and configure the data node availability.

This release comes with a wealth of new capabilities and performance enhancements, such as

• Improved sort performance and compression
• Expressive developer features
• More options for logical replication
• New public schema permissions.

Learn more on the release in PostgreSQL release notes. For information on Aiven for PostgreSQL, see the service’s documentation.

• Prometheus query builder: allows you to search and select multiple metrics and labels simultaneously.
• A new heatmap model: that is highly performant and orders of magnitudes faster.
• Improved User Interface: enhanced Dashboard previews that provide a graphical representation of your dashboard to differentiate between multiple dashboards easily. The Dashboard previews feature is available as an opt-in feature in the Aiven Console under Advanced Configuration for your Aiven for Grafana service.

For more information, see Aiven for Grafana.