date: 2026-04-21
title: 'Spice v2.0-rc.3 (Apr 21, 2026)'
type: blog
authors: [ewgenius]
tags: [release, data-connector, databricks, snowflake, adbc, kafka, opentelemetry, observability]
Announcing the release of Spice v2.0-rc.3! ⚡
v2.0.0-rc.3 is the third release candidate for advanced testing of v2.0, building on v2.0.0-rc.2.
Highlights in this release candidate include:
- HTTP Connector Enhancements with OAuth2 refresh-token authentication, query-parameter pagination, and map-to-array conversion for broader API compatibility
- Databricks and Unity Catalog Reliability Improvements with resilience controls, improved UC-awareness, permission checks, and structured error reporting
- Snowflake and ADBC Registration Performance Improvements with better observability during dataset registration
- OpenTelemetry Exporter Improvements with exporter fixes and support for authenticated metrics export
- Kafka, GitHub, and HTTP Connector Fixes including Kafka reliability improvements, GitHub GraphQL resilience updates, and HTTP JSON union/reload fixes
What's New in v2.0.0-rc.3
HTTP Connector Enhancements
The HTTP connector now supports more authentication and API response patterns, making it easier to integrate with modern REST APIs.
Key improvements:
- OAuth2 Refresh-Token Authentication: Added support for OAuth2 refresh-token flows for APIs that issue short-lived access tokens.
- Query-Parameter Pagination: Added pagination support using query parameters for APIs that expose page or cursor controls in the URL.
- Map-to-Array Conversion: Added response transformation support for APIs that return map-shaped payloads that need to be normalized into arrays.
- Improved JSON Union Handling: Better handling for heterogeneous JSON payloads during ingestion.
- More Reliable Reloads: Fixed runtime behavior for HTTP-backed datasets during spicepod reloads.
Example configuration of an HTTP connector using the OAuth2 refresh token flow:
Databricks and Unity Catalog Reliability Improvements
Databricks and Unity Catalog integrations are now more resilient and provide clearer behavior in permission-constrained environments.
Key improvements:
- Resilience Controls: Added controls to improve reliability when interacting with Databricks services.
- Unity Catalog Awareness: Improved handling for Unity Catalog-specific behaviors and mixed deployment configurations.
- Permission Prechecks: Databricks UC permission checks now distinguish explicit denials from ambiguous cases.
- Structured Error Reporting: Advisory permission failures now surface with more actionable structured errors.
- Classic SQL Warehouse Compatibility: Improved handling for foreign table scenarios with Classic SQL Warehouse combinations.
- Task History Instrumentation: Added instrumentation to improve observability for Databricks-related operations.
Snowflake and ADBC Improvements
Snowflake and ADBC-backed dataset registration is now faster and easier to observe.
Key improvements:
- Faster Dataset Registration: Improved registration performance for Snowflake and ADBC datasets.
- Better Observability: Added better instrumentation and visibility into registration workflows.
- ADBC Alignment: Updated ADBC dependencies and integration points for improved compatibility.
- Search Schema Fix: Fixed a full-text search schema mismatch issue with the ADBC connector.
OpenTelemetry and Observability Improvements
Spice improves telemetry export reliability and authenticated metrics delivery for the OpenTelemetry integration.
Key improvements:
- OTEL Exporter Fixes: Fixed issues in the OpenTelemetry exporter.
- Authenticated Metrics Export: Added support for authorization headers in the OTEL metrics exporter.
- Reduced Startup Noise: Suppressed unnecessary AWS SDK noise and improved OTEL-related initialization behavior.
- Connector Initialization Reliability: Fixed issues that could block connector initialization in telemetry-related code paths.
Dependency and Toolchain Updates
| Dependency / Component | Version / Update |
|---|
| Rust toolchain | v1.94.1 (from v1.93.1) |
| DataFusion | v52.5.0-rc1 |
| mistral.rs / candle | mistral.rs v0.8.x, candle v0.10.1 |
| ADBC Core | v0.23 |
Other Improvements
- Improved Query Pushdown: Expanded sort and limit pushdown, including improved pushdown behavior for Oracle and MSSQL connectors.
- Partitioned Query Planning Improvements: Improved
PartitionedTableScanRewrite handling for ORDER BY, partition expressions, and fully qualified table references, while preventing incorrect bucketing partition pushdown to executors.
- MongoDB SRV Support: Upgraded
datafusion-table-providers with MongoDB SRV support.
- Tantivy Logging: Search logging now defaults to
warn unless very verbose logging is enabled.
- Kafka Connector Fixes: Improved reliability for the Kafka connector behavior.
- GitHub Connector Resilience: Improved commit fetching for dynamic and slash refs, and reduced GraphQL page sizes on gateway errors for the GitHub connector.
- GitHub API Efficiency: Lowered default comment fetch sizes to reduce pressure on GitHub GraphQL APIs.
- Embedding Validation: Added validation for embedding
row_id columns during dataset initialization.
Contributors
Breaking Changes
No breaking changes.
Cookbook Updates
No new cookbook recipes.
The Spice Cookbook includes 86 recipes to help you get started with Spice quickly and easily.
Upgrading
To upgrade to v2.0.0-rc.3, use one of the following methods:
CLI:
Homebrew:
Docker:
Pull the spiceai/spiceai:2.0.0-rc.3 image:
For available tags, see DockerHub.
Helm:
AWS Marketplace:
Spice is available in the AWS Marketplace.
What's Changed
Changelog
- fix: Full Text Search schema mismatch with ADBC connector by @lukekim in #10235
- docs: Update v2.0.0-rc.2 release notes with latest changes by @lukekim in #10238
- Fix append refresh dedup failure when refresh_sql selects column subset by @sgrebnov in #10225
- Revert "Properly mark dataset as Ready on Scheduler (#10215)" by @sgrebnov in #10242
- Fix failing merge conflicts for benchmarks by @krinart in #10247
- fix(github): fetch commits for dynamic and slash refs by @lukekim in #10233
- Upgrade DataFusion to v52.5.0-rc1 by @lukekim in #10249
- Merge develop to trunk (2026-04-09) by @claudespice in #10248
- fix: Validate embedding row_id columns during dataset init (fixes #8226) by @claudespice in #10208
Full Changelog: https://github.com/spiceai/spiceai/compare/v2.0.0-rc.2...v2.0.0-rc.3
