dataaaaa!

Kestra 2.0 expands plugins beyond integrations to platform internals, adding custom policy rules, file renderers, and log data stores. Notably, flow-level pluginDefaults is removed, and open source now supports external log data stores to streamline migrations.

Kestra 2.0 introduces Cases to embed incident management directly in the orchestrator. It dedupes repeated workflow failures into a single record with SLA tracking, remediation flows, and structured resolution reasons, eliminating external incident sync tools.

Kestra 2.0 orchestrates workflows using declarative YAML over existing code. Its new Controller decouples workers from the database, letting them run in isolated networks via outbound gRPC without inbound ports. This architecture avoids direct database credentials on workers.

A Kestra meetup in Paris brings together practitioners to discuss building and shipping AI products in production. Industry leads from Sanofi and RTK-AI Labs join Kestra product management for a practical roundtable covering the architectural and data engineering realities of operationalizing AI.

Kestra 2.0 decouples its control and data planes, allowing workers to run anywhere via outbound-only gRPC streams without direct database access. The engine also decouples the queue from the repository, unifying the architecture while supporting backends like Postgres, Kafka, Redis, or AMQP.

Kestra 2.0 introduces a unified orchestration engine with decoupled queues and repositories. Workers now run anywhere using outbound gRPC with no database access. Data engineers can keep Postgres and switch queues to Kafka or Redis without touching flows, keeping execution next to the data.

Kestra 2.0 rebuilds orchestration with separated control and data planes. Workflows register as tools on MCP servers, while a revised Loop task executes iterations as isolated sub-executions to prevent executor memory exhaustion. Teams manage pipeline drafts safely without impacting production runs.

Kestra 2.0 rebuilds its execution engine, doubling sustained throughput to 4,000 executions per minute on simple flows. A streamlined single-consumer queuing mechanism and decoupled loop sub-executions significantly reduce database trips, message payload sizes, and executor CPU overhead.

Kestra 2.0 lets data engineers expose workflows as agent tools via an MCP trigger. Agents invoke flows using generated JSON schemas, triggering tracked executions with full audit logs. Kestra treats those calls exactly like every other execution in the system, maintaining existing RBAC and retries.

Kestra 2.0 decouples queue and repository backends under a unified engine. Data engineers can run Postgres for both, add AMQP/Redis for lower latency, or use Kafka for horizontal scale paired with Postgres or Elasticsearch, eliminating the operational complexity of Kafka Streams.

Kestra optimized its data plugins by removing repetitive hot-path operations. Key changes include enabling JDBC connection pooling by default and caching ResultSet metadata once per query instead of per row, cutting overhead and garbage collection.

Kestra 2.0 rewrote its orchestration engine to decouple workers from the database via a new Controller service over gRPC. It unbundles the queue and repository, letting engineers choose backends like Kafka, Redis, or JDBC independently to eliminate database bottlenecks.

Kestra released a DigitalOcean plugin enabling declarative orchestration for Droplets, DOKS, and managed databases like Kafka and Postgres. It automates ephemeral compute pipelines and teardowns, letting engineers orchestrate batch infrastructure with secure secret handling and state polling.

Enterprise data pipelines combine Kestra, dlt, and Snowflake to separate orchestration from business logic. Using Kestra's SyncNamespaceFiles plugin, engineers pull code bases from GitHub into namespace files. The project demonstrates a real-world data pipeline using Kestra as the orchestrator.

Kestra 2.0 re-engineers its open-source execution engine for scale and distributed pipelines. Stateless workers communicate with the control plane via gRPC without direct database access, and large task outputs load on demand to preserve database performance across thousands of tasks.

Kestra orchestrates hardware EDA workflows by managing containerized RTL verification and synthesis. Using tools like Verilator and Yosys, pipelines run parallel test suites, extract gate-count metrics, and enforce automated quality gates that fail executions exceeding silicon budgets.

Kestra built standalone shell scripts to automate local dev stacks, shadow JAR builds, and fleet management across 200+ plugin repos. These commands provide a single execution contract for both engineers and AI agents, cutting token costs and eliminating repetitive manual CLI tasks.

Kestra decouples orchestration logic from task execution. The executor walks the DAG and dispatches tasks to an internal queue without external calls. Workers independently poll the queue to run scripts, APIs, or queries, letting teams scale task compute across Kubernetes or isolated worker groups.

Scaling orchestration requires separating execution, isolation, and workload routing. Kestra achieves this via an active-active architecture on Kubernetes, hierarchical namespaces and tenants for RBAC, and worker groups that route tasks across clouds or hybrid clusters.

Kestra bridges the gap between agent reasoning frameworks and production engineering. By wrapping AI agent tasks in declarative YAML workflows, it adds critical operational guardrails: automated retries, timeouts, event-driven scheduling, secret management, and human-in-the-loop approvals.

Kestra automates its plugin development lifecycle by using context engineering instead of simple prompting. Agents build, test, and review plugins within structured environments while human checkpoints maintain code quality. Teams execute tasks faster by eliminating repetitive manual boilerplate.

Kestra unified its ecosystem release process by requiring deliberate version tags instead of automated merges. Pushing a tag triggers builds via native tools like GoReleaser, PyPI, and Maven Central, ensuring consistent deployment across SDKs, the Terraform provider, and CLI tools.

Kestra orchestrates end-to-end AI pipelines, managing document parsing, embedding generation, and vector indexing. It provides integrations for vector stores like Pinecone and Weaviate, supporting retry mechanisms, bulk vector ingestion, and automated change-stream re-indexing.

Kestra automates RAG workflows by turning indexing and retrieval pipelines into declarative YAML. It provides scheduling, retries, and observability for document chunking, embeddings, and vector storage, supporting stores from built-in KV to Qdrant or PGVector.

Kestra automated its multi-version release pipeline by replacing manual GitHub UI checks with Kestra Flows. Using LoopUntil tasks and the GitHub CLI, the orchestrated workflow polls CI status and handles retries, freeing engineers from monitoring up to eight parallel CI jobs across repositories.