Skip to content

2610 Update

Latest

Choose a tag to compare

@joshgrayscale joshgrayscale released this 11 Nov 23:55
0708709

Azure IoT Operations 2610 (v1.5.45) public release notes

Release date: 11 November 2026
Release type: Milestone (GA)
GA version: 2610 (v1.5.45) · Version history
Azure Government version: 2610 (v1.5.46), public preview

Azure IoT Operations 2610 (v1.5.45) introduces the Essentials SKU, a new lower-cost pricing option for single-node deployments. It also brings SQL Server data into your data flows and extends management actions to HTTP/REST assets. Azure IoT Operations is now available in public preview in Azure Government. Improvements across data processing, OPC UA connectivity, MQTT messaging, and upgrade recovery help keep deployments running consistently.

Release highlights

  • Azure IoT Operations Essentials SKU: Connect assets and process their data with a new lower-cost option for single-node deployments. Essentials includes the MQTT broker, Akri and data flows; choose Standard for OPC UA connectivity or multi-node deployments. For SKU pricing, see Azure IoT Operations pricing.
  • Azure Government public preview: Deploy Azure IoT Operations in US Gov Virginia with the 2610 GA feature set. See Azure Government (public preview) in these notes for version and availability details.
  • SQL Server connectivity: Bring valuable context from SQL Server into your data flows, with datasets defined by SQL queries.
  • HTTP/REST management actions: Configure management actions for assets connected through the HTTP/REST connector.
  • Updated navigation: Try a new way to find your instances and resources in the operations experience. Switch between the new and existing layouts to explore the changes at your own pace.
  • Custom graph processing: Use custom WebAssembly modules in data flow graphs to call request-response services over MQTT for data enrichment, lookups, routing decisions, and asset management actions.
  • Data processing improvements: Reduce memory use when writing to Delta Lake, and improve processing reliability with fixes to shared enrichment, MQTT source sessions, and batch acknowledgments.
  • OPC UA reliability: Avoid telemetry interruptions caused by trust-list configuration, and get clearer asset status when monitored-item requests fail.
  • Media processing: Extract media based on event time to local-shared and cloud-ingest volumes, and manage retention for local storage and local shared volumes.
  • Upgrade recovery and SKU consistency: Retry failed upgrades without being blocked by stale state. Instance updates preserve the selected SKU, including updates made through older API versions.
  • AI-assisted data flow authoring in VS Code: Describe your requirements in natural language to generate data flow configurations with GitHub Copilot, validate them locally, and deploy them to your Azure IoT Operations instance.
  • View live data in VS Code: Inspect live MQTT topics and messages from your Azure IoT Operations broker, using topic filters to focus troubleshooting on relevant traffic.
  • Deployment Image List (DIL): Download the DIL for this release to support dependency scanning and private-registry mirroring. Download the DIL for this release | How to mirror images to a private registry

Upgrade recommended. This release includes important improvements for customers using Delta Lake or Parquet destinations, shared state store enrichment, persistent MQTT source sessions, OPC UA, or SQL Server connectors. It also includes updated certificates and dependencies, described under Security.

Upgrade considerations

Review these changes before updating your deployment:

  • Early 2610 deployments: If you deployed 2610 version 1.5.41, update to 1.5.45. Version 1.5.41 doesn't include the certificate and dependency updates in this release.
  • Essentials SKU: The Essentials SKU is available for new instances. Existing Standard SKU instances, including instances upgraded to 2610, remain on the Standard SKU and can't be changed to Essentials. To use the Essentials SKU, deploy a new instance.
  • Broker metrics export: Broker metrics now export directly to OpenTelemetry Protocol (OTLP) gRPC collectors. For Prometheus scraping, configure a Prometheus exporter on the collector. The former broker diagnostics-service endpoint is removed. The spec.diagnostics.metrics.prometheusPort setting remains for compatibility but has no effect.
  • Delta Lake table maintenance: Configure external retention and cleanup for Delta tables written by data flows. Data flows continue to append data and create checkpoints but no longer clean up expired transaction logs. Table owners are responsible for retention and cleanup.
  • OPC UA message sequence numbers: Sequence numbers are now maintained independently for each dataset or event group. Update consumers that assume a single asset-wide sequence when detecting missing messages or connector restarts.
  • Cloud destination addresses: Azure Data Lake Storage Gen2, Azure Data Explorer, and Microsoft Fabric OneLake endpoints now enforce service-specific HTTPS address validation. Documented bare hostnames remain supported. Correct addresses containing an unsupported scheme, embedded credentials, extra paths, or a hostname for the wrong service before applying an update.

Version support: Azure IoT Operations follows the N-2 version support policy. Check the version history and support documentation when planning your upgrade.

Known issues

For active known issues and workarounds, see Known issues.

  • ONVIF discovery: Discovery fails when more than 10 media endpoints are detected.

Azure Government (public preview)

Azure IoT Operations is available in public preview in Azure Government, in the US Gov Virginia region.

  • Version: Azure Government uses 2610 (v1.5.46), which delivers the 2610 GA feature set with updates specific to Azure Government. The Azure IoT Operations CLI selects this version when you deploy to Azure Government.
  • Deployment Image List (DIL): Download the Azure Government DIL for v1.5.46.
  • Deployment and management: Use Azure CLI and the operations experience. Azure portal support for Azure IoT Operations isn't available yet in Azure Government.
  • Not available: Preview features and the preview release channel (including CLI preview options), features that depend on Azure Container Storage enabled by Azure Arc, and legacy assets and asset endpoint profiles.
  • Data flow destinations: Use cloud endpoints in Azure Government.

For details, see Azure IoT Operations in Azure Government.

Platform

New features

  • Azure IoT Operations Essentials SKU: Choose a new lower-cost pricing SKU for single-node deployments that need the MQTT broker and data flows without the OPC UA connector. The current Standard SKU supports single-node and multi-node deployments and includes OPC UA connectivity. Select the SKU when you create an instance; existing Standard SKU instances can't be changed to Essentials.

Fixes

  • SKU preservation on update: Updates that omit the SKU preserve the existing selection, including updates made through older API versions, so updates aren't rejected as unintended SKU changes.
  • Default SKU consistency: New instances created without an explicit SKU consistently use the Standard SKU.
  • Recovery after failed upgrades: A terminal upgrade failure clears the in-progress marker and records the failure, allowing a later retry to proceed.

Azure portal

New features

  • Instance SKU selection: Choose the Essentials or Standard SKU in the deployment wizard to match your cluster and connectivity requirements.
  • Deployment release channel: Choose a GA or preview deployment. The wizard generates Azure CLI commands for the selected release channel.

Improvements

  • Compatible connector versions: Find Microsoft-provided connector versions that match your instance’s release channel. GA instances show compatible GA versions; preview instances can also show compatible preview versions. This filtering does not change connector behavior for customer-managed registries.

Azure IoT Operations experience

New features

  • SQL Server connectivity: Bring SQL Server data into your data flows. Configure a SQL Server connection and associated assets, then define datasets using SQL queries.
  • GA and preview experiences: Explore new capabilities on instances running preview versions alongside your GA instances. Available connectors and transforms match each instance's version.

Improvements

  • Updated navigation: Try a new way to find your instances and resources. Switch between the new and existing layouts to explore the changes at your own pace.

Fixes

  • Built-in MQTT broker setup: Configure devices to use the built-in MQTT broker without selecting external-broker authentication options. Authentication requirements for external brokers are unchanged.

Azure IoT Operations CLI

New features

  • Instance SKU selection: Select the Essentials SKU when creating an instance through Azure CLI deployment workflows.

For the full changelog, see Azure IoT Operations CLI releases.

Connectors

SQL Server

New features

  • SQL Server data acquisition: Deploy an ODBC-based connector to collect SQL Server data for Azure IoT Operations.

OPC UA

New features

  • Dataset heartbeats: Configure the OPC UA connector to write a dataset heartbeat counter to the OPC UA server so the server can monitor connector liveness.
  • Opt-in delivery acknowledgments: Configure the OPC UA connector to write a delivery acknowledgment counter to the OPC UA server after a dataset or event is fully processed and successfully published to the Azure IoT Operations MQTT broker.
  • Data-change filters: Configure when the OPC UA server reports data point changes using OPC UA data-change filters.
  • Data-quality filters: Configure the OPC UA connector to exclude data point changes whose OPC UA status code isn't Good.
  • Cyclic reads: Configure the OPC UA connector to read a dataset at a fixed interval.
  • On-demand keyframes: Request a keyframe through an MQTT remote procedure call (RPC).

Improvements

  • Opt-in management action retries: Stateful execution can retry eligible management actions after a reconnect while respecting the remaining execution deadline. Non-idempotent operations are not replayed after an ambiguous failure.
  • Opt-in immediate failover: Configure the OPC UA connector to transfer leadership to the passive instance as soon as a session failure triggers reconnection.
  • Trigger-based keyframes: Configure the OPC UA connector to generate a keyframe when a designated triggering data point changes, including cached values for unchanged data points.

Fixes

  • Connector enablement and disablement: Enabling or disabling OPC UA keeps connector-template provisioning coordinated with the connector lifecycle, preventing missing resources and stalled template operations.
  • Trust-list compatibility: Devices with a trust list but no trust-mode property no longer cause repeated supervisor restarts or interrupt telemetry and upgrades.
  • Configuration validation: Property names and enum values are validated consistently with the connector’s case-insensitive runtime parsing.
  • Monitored-item errors: Failed monitored-item requests now report the request-level error in asset status instead of crashing the connector when per-item errors are unavailable.
  • Authenticated asset management: Asset materialization using username-and-password authentication avoids conflicting service-client sessions that could cause timeouts.
  • Per-stream sequence numbers: Each dataset and event group has its own message sequence, supporting gap and restart detection for consumers subscribing to only part of an asset’s data.
  • Heartbeat metric export: OPC UA heartbeat metrics no longer contain duplicate attributes that prevent export through Prometheus-compatible observability pipelines.

Breaking changes

  • Reduced MQTT message metadata: The OPC UA connector now includes fewer MQTT user properties by default, reducing message header size. Consumers that depend on previously included properties must configure opcUaConformance and messageDiagnostics to restore the previous behavior.

For configuration guidance, see OPC UA connector.

MQTT

Fixes

  • Request-response correlation: Connector runtime schema and state requests keep late responses associated with their original request, preventing a timed-out request from returning the wrong result to a later caller.

HTTP/REST

New features

  • Management actions: Configure management actions for assets connected through the HTTP/REST connector.

For configuration guidance, see Management actions.

Data flows and data flow graphs

New features

  • Service calls from custom graph modules: Enrich messages, look up data, and make routing decisions by invoking request-response services over MQTT from custom WebAssembly (WASM) modules in a data flow graph. The mRPC command invoker host API returns the correlated response or an invocation error.
  • AI-assisted data flow authoring in VS Code: Use the Azure IoT Operations Data Flow VS Code extension's Copilot Chat participant to create data flows for transformation, filtering, windowing, and state store enrichment using natural language. Generate configurations grounded in resources from the connected instance, create sample input data, and validate behavior locally before saving or deploying.
  • View live data in VS Code: Start a live data session from the Azure IoT Operations Data Flow VS Code extension to inspect live MQTT topics and messages. Configure topic filters to narrow the traffic shown while troubleshooting data flows.

Improvements

  • Delta Lake append performance: Append workloads retain less unnecessary file metadata, and checkpoints run in the background at configured intervals. These changes reduce memory growth and transaction-log overhead. Retention and expired-log cleanup are now managed externally by table owners.

Fixes

  • MQTT source session cleanup: Removing a source or scaling down a data flow cleans up its persistent broker sessions. Stale subscriptions and queued messages no longer divert shared-subscription traffic from active instances.
  • Shared state store enrichment: Multiple data flows subscribing to the same enrichment dataset receive initial data and subsequent updates instead of waiting indefinitely or using stale enrichment.
  • Batch acknowledgment handling: A failed record commit preserves acknowledgment tracking for the accumulated output batch, allowing subsequent source acknowledgments to progress.
  • Schema and state request timeouts: Late responses to timed-out requests no longer satisfy a later request for a different schema or state value.
  • Operator startup: The data flow operator skips unchanged custom-resource conversion settings, avoiding unnecessary Kubernetes API updates that could block startup and connector creation. Certificate and endpoint changes continue to be applied.
  • Unsupported Parquet and Delta schemas: Unsupported field types and malformed schemas produce configuration or processing errors instead of crashing schema setup. This change does not add support for new field types.
  • Cloud endpoint validation: Azure Data Lake Storage Gen2, Azure Data Explorer, and Microsoft Fabric OneLake destinations accept documented hostnames and valid HTTPS origins while rejecting unsupported or misleading addresses.
  • MQTT message-rate reporting: Source and destination statistics account for idle connections, avoiding misleading rates and missing reports after bursts of traffic.
  • WASM controller resource cleanup: Completed controller requests release their worker threads, preventing thread and memory accumulation.
  • WASM artifact download timeouts: Overall time limits prevent slow OCI artifact transfers from blocking subsequent downloads indefinitely.
  • Signed artifact downloads: Valid PEM certificates in Kubernetes Secrets are read without an extra base64 decode, allowing signed graphs and modules to download.
  • OpenShift WASM cache access: WASM artifact caching works under the OpenShift restricted-v2 security context.

For configuration guidance, see Data flows.

Messaging and MQTT

New features

  • State store queries: Enumerate keys in pages with SCAN and read multiple keys with MGET. These operations are supported by the broker; availability in an application depends on its client SDK.

Fixes

  • Recovered state expiry: State store keys restored after recovery expire according to their time-to-live settings instead of remaining readable after expiration.
  • Persistent-session authorization: Resuming a persistent MQTT session rechecks authorization for its existing subscriptions. Reconnection is rejected when those subscriptions are no longer authorized.
  • Subscription recovery: Broker recovery prevents duplicate subscriptions after a restart.
  • Authentication responses: AUTH and successful CONNACK responses include the required authentication method, improving interoperability during connection establishment and reauthentication.
  • MQTT client error reporting: Canceled operations retain their cancellation reason instead of being reported as detached operations.
  • Invalid authentication packets: Non-compliant authentication packets return a protocol error instead of causing the MQTT client to panic.

Observability

Improvements

  • Broker metrics export: Send broker metrics directly to an OTLP gRPC collector. Prometheus scraping is supported through a Prometheus exporter configured on the collector. See Upgrade considerations for the removed diagnostics endpoint and configuration changes.

Fixes

  • Broker health diagnostics: MQTT diagnostics retry transient connection failures, report missing observations as failures, and avoid counting dependent failures repeatedly. Latency metrics reflect successful operations rather than rejected acknowledgments or stale healthy results.

Security

Fixes

  • Certificate and dependency updates: Azure IoT Operations 2610 (v1.5.45) includes replacement certificates and rebuilt components and dependencies affected by a revoked third-party intermediate certificate authority.

For more information, see the Azure IoT Operations documentation.