Enhanced High Availability to allow for synchronizing both project configuration and state between one or more Intelligence Hubs. The active hub starts with the latest configuration and state.
Added ‘For Each’ and ‘While’ loop stages to Pipelines to iterate over each element in an array while preserving the original order.
Added a new Dashboard page that provides an overview of the configuration, the status of connections and pipelines, and quick visibility to the latest event log messages.
Removed the need to login when opening new tabs. Login sessions are now persisted up to the configurable session timeout.
Added a new MCP Services page to the main navigation to view the MCP tools surfaced by Pipelines and Connections.
Added support for new MCP tools for Pipeline Configuration that allow AI Agents to create and edit pipelines.
Added a new MCP Client connection to aggregate tools from other MCP servers.
Added out-of-the-box MCP tools to the OPC UA TCP and MQTT Connections for browsing and reading data. These tools are enabled in the Connection settings.
Added support for creating Models from Connection Inputs and manual entry of JSON payloads. This allows quick creation of Models from SQL tables, for example.
Added support for importing OPC UA Nodesets as Models.
Improved the performance of the modeling stage by as much as 10x for some use cases.
Added Microsoft Fabric OneLake Connection with Output support. This connection can be used to send files to an Azure Data Lake Storage Gen2 (ADLS2) container.
Added Input support for Azure Blob Storage connection.
Added Databricks SQL Connection with Input support. This connection can query tables within Databricks.
Added a new JSONata transform stage for querying and transforming JSON values. The former “Transform” stage has been renamed to “JavaScript”.
Added Delay stage to Pipelines.
Added support for clearing out Store-and-Forward files through the UI, allowing users to recover disk space.
Enhanced the PostgreSQL Connection to support writing to partitioned tables.
Added pipeline statistics for the number of events per second that a pipeline is processing.
Added support for browsing Namespaces while building a Smart Query stage.
Enhanced the Breakup Stage to allow breaking up to a certain depth, no longer requiring multiple breakup stages to process hierarchy.
Added the ability to see and drag and drop Input and Instance parameters from the reference panel into the Input and Instance configuration. Removed parameter summary from Inputs and Instances as part of this change.
Enhanced the reference UI in Instance Attribute and Pipeline Stages to allow for copy & paste and manual editing of references.
Added audit logging support for Project Sync actions and other bulk edits.
Added Keep-Alive support for SQL Connections with Change Data Capture (CDC) enabled.
Added Connect and Session timeout to the File Connection settings when using SFTP.
Added a new Pending Write Size connection statistic that represents the number of pending store and forward writes.
Added a new Browse input type to the MQTT Connection.
Enhanced Oracle CDC with the ability to read from multiple tables, support for reading CDC records from Pluggable Databases (PDB) within multitenant databases, and a configuration option to control the starting timestamp for the first CDC read.
Added a new setting to PI System Event Frame inputs to control if the read returns child event frames.
Added query paging support to the PI Asset, Asset Metadata, and Event Frame input types. This allows for queries to be split up to improve performance.
Added support for a new path-based Identifier Mode for the OPC UA Connection. In this mode, the input stores a relative path (using OPC Qualifier Names) that can be used to translate to Node IDs at runtime.
Added hyperlink to the connection dropdown on the Write New Stage in pipelines.
Improved links to open more consistently in new tabs.
Changed the Application Settings page layout to be more condensed and support collapse/expand.
Added support for renaming Certificates after importing.
Updated the toast card styling and added dark mode support.
Standardized tooltip styling throughout the product to be left-aligned.
Standardized tags to be sorted alphabetically throughout the product.
Added log messages for failures when saving configuration files or backups.
Added support for array type parameters to the API Trigger.
Changed the default REST Data request timeout from 30 seconds to 10 minutes.
Fixes:
Fixed an issue with the request timeout not being properly applied to the PI System AF SDK Agent.
Fixed an issue with Google Cloud Pub/Sub Store-and-Forward occasionally dropping messages during connection loss. Google Cloud Pub/Sub Connection now operates on an “at least once” standard. All messages will be sent and received at least once and may be sent more than once.
Fixed an issue where a value of quality Good No Data reflecting as ‘Failed’ in the input test reader to correctly reflect a ‘Success’ status.
Fixed an issue where the user-supplied name and email for Git backup commits were ignored.
Fixed an issue that caused Smart Query with no results to return an error instead of Good No Data.
Fixed an issue where Amazon Kinesis Data Streams and Amazon Data Firehose connections failed to complete Store-and-Forward writes after a connection loss.
Fixed an issue where buffer flushes could bypass the trigger queue limit for Pipelines.
Fixed and issue with Oracle Query and CDC inputs having incorrect or missing data with some datatypes.
Fixed an issue where high frequency debug stage events would be returned in the opposite order when viewing a historical replay or debug event.
Fixed an issue with the Apache Parquet connection where some Oracle CDC inputs would cause an error.
Fixed an issue causing Server Message Block (SMB) directory inputs to be slow.
Fixed an issue that caused the remote configuration client to generate a new client ID each time the application was restarted instead of using the existing ID.
Fixed an issue where Sparkplug Birth Certificate Metrics without timestamps would fail to be read. Missing metric timestamps now fall back to the payload’s timestamp or to the Unix Epoch (January 1, 1970) if the payload timestamp is also missing.
Fixed an issue with Sparkplug where group level subscriptions could cause timeout sending birth requests.
Fixed an issue that would cause an ‘Unknown Source’ error when attempting to batch updates for the AWS IoT SiteWise API.
Improved the error message when a Snowflake Streaming Connection is missing or has an invalid certificate.
Fixed an issue where the Write and Write New stages would wait for a pipeline event to be queued when leveraging the Ignore Return setting, causing delays in Pipeline event processing.
Fixed an issue where PI writes could return “point already exists” errors when create was enabled and the case of the PI point name changed.
Fixed an issue in the Sparkplug Connector that prevented the model name from being populated on Metrics properties that were Templates.
Fixed an issue where setting and removing the tag filter in the UI would show all nested groups and values at the same level.
Fixed an issue where Snowflake Streaming channel names created a GUID that could lead to too many channels being created for a table. Channels names now use the hostname, connection name, and table name to guarantee uniqueness per hub.
Fixed an issue where the error count for the Write stages would increment twice when using the “Ignore Return” value.
Breaking Changes:
Google Cloud Pub/Sub connections now operate on “at least once” logic, so messages may be sent more than once in connection-loss scenarios to guarantee delivery.
Replaced sqlite-jdbc-crypt with the xerial/sql-jdbc driver to enable better interoperability of the HighByte event/audit logs and keep the driver better up-to-date with the latest security patches. Log entries are now signed by the application. Entries that have been modified outside the application are marked as such when viewing logs through the UI.
Added verification button to the Event Log Page to check the authenticity of all logs in the log databases.
Security Patch Updates:
Runtime:
CVE-2025-48924: Defect that can cause stack overflow and crash when resolving classes.
CVE-2024-52338: Deserialization flaw that could allow code execution when reading untrusted data.
CVE-2025-55163: HTTP/2 processing issue that could enable denial of service.
CVE-2025-24970: Defect that allows a crafted SSL packet to cause a native crash in SslHandler.
CVE-2025-58056: Defect that permits HTTP request smuggling via incorrect parsing of chunk extensions.
Frontend:
CVE-2025-7783: Weak randomness in form boundaries that could allow parameter pollution.
CVE-2025-58751: Defect where files with names like those in the public directory are served bypassing access control.
CVE-2025-58752: Defect that serves HTML files outside the intended directory, ignoring server.fs restrictions.
CVE-2025-58754: Defect that allows a data: URL with a large payload to exhaust memory and crash the process.
Azure Blob Storage:
CVE-2025-22227: Redirect handling flaw that could expose credentials.
Ignition Module:
CVE-2024-30171: Timing side channel in TLS handshake that could leak information.
CVE-2024-29857: Curve parameter validation issue that could cause denial of service.
CVE-2023-33201: LDAP injection vulnerability during certificate validation.
