• Fixed an issue where an OPC UA Connection could consume too many threads in certain error conditions.
• Fixed issues where invalid Sparkplug metrics could cause read failures for the entire Input.
• Fixed an issue where Webhook Inputs were incorrectly parsing XML arrays into strings.
• Fixed an issue with the UNS Client where subscribing to an invalid topic caused rapid re-connection attempts.
• Fixed an issue where browsing an empty OPC UA branch navigated the user to an empty screen with no tags to select. The user now stays on the same page with a message saying no tags were found.
• Fixed an issue where the Flow Trigger Delay was not shown for Flows that were in Polled mode.
• Fixed an issue where enabling the “Ignore Self Signed Certificates” for MQTT and Sparkplug Connections disabled the ability to use Client Certificate/Key authentication.
• Fixed an issue where refreshing the Connection Statistics page caused outputs to be de-selected in the UI.
• Removed a warning produced for SQL Output updates in cases where no rows were found to update.
• Fixed an issue where switching from a PI System Input Point Current Value to a Raw Values read could cause read errors.
• Fixed a UI issue where additional read sources could be added to a Read Stage when a Pipeline was in replay mode.
• Fixed an alignment issue in the OPC UA Input Methods UI.
• Fixed a bug where incorrect text was shown over the Tag selector when hovering.
• Fixed an issue where a plus icon was shown in the SQL table browser.
• Fixed an issue where Project page step numbers had incorrect formatting.
• Fixed an issue where console.log messages logged in Pipeline Transform stages were not shown in the UI for troubleshooting.
• Fixed an issue with the Docker image where the Store and Forward directory did not default to the appData directory, resulting in Store & Forward data being lost on container reboots.
• When “Include Metadata” is enabled, REST Client reads that return non-200 level responses now return the error code and message in the same format as 200 responses.
• Fixed an issue where the UNS Client could not connect to secure MQTT connections.
• Fixed an issue where SQL Inputs that did not run SELECT queries returned bad status.

 Security Patch Updates:

• Amazon Kinesis Data Firehose
  • CVE-2024-21634: DOS when deserializing data from untrusted source. It is not possible to execute this code in the HighByte Intelligence Hub implementation.

​Patch (3.4.4 2024.6.12.2):

• Added support for Sparkplug Inputs to receive data from edge nodes that do not publish birth certificates. This does not work if metrics use aliases in the publish.
• Added Binary and ASCII modes to the File FTP Connection to support reading/writing binary files without modifying them. 
• Changed the UI to hide password fields when they are being entered in connections settings.​
• Improved Snowflake Streaming output performance by avoiding metadata queries that caused warehouse compute costs. 
• Fixed a bug where the product name was changed from “HighByte” to “Highbyte”. This could result in existing OPC UA Connections failing with a Bad_CertificateUriInvalid error, because the name in the certificate did not match the name used in the connection.
• Fixed a bug with SQL Outputs where column names were being matched to the output data as case sensitive, resulting in failed writes or missing data.
• Fixed issues with OPC UA reading empty primitive arrays and null objects in Structured Data types.
• Fixed an exception thrown and displayed in the logs when writing to an invalid IP or hostname from a Write New stage.
• Fixed an issue where remote hubs with older versions could connect to a central hub on version 3.4 but configuration requests would timeout.
• Fixed an issue where SQL Inputs with multiple lines returned the first line and not the first result. For example, if a statement had an insert followed by a select, it would return “good no data”. Now it returns the select result.
• Fixed an issue where templated Inputs or Instances that had a dynamic reference that required more than one read could cause a deadlock. An example would be a Condition that parses SQL data into template parameters. The deadlock could occur on startup or anytime the runtime is restarted (e.g., settings change, redundancy change over).
• Fixed an issue where SQL connections would not recover from a disconnect in some cases.
• Fixed OPC UA Connection threading issue that could result in reads and writes no longer working on the connection. This could occur in some cases where the connection is dropped because of request timeouts or other network issues. 

Patch (3.4.5 2024.7.18.2):

• Fixed a bug with Sparkplug CMD inputs where multiple commands sent in a single message could cause event based Flows to not execute. 
• Fixed an issue with Microsoft SQL Server, PostgreSQL, Google Big Query inputs where reads would check the SQLite cache for the latest index value when indexing was disabled. 
• Fixed an issue with Active Directory (AD) authentication where a blank password could grant product access if the ‘LDAP unauthenticated authentication’ isn’t explicitly disabled in AD. Logins are now rejected if the username or password fields are empty.
  

Patch (3.4.6 2024.11.4.2) 

• Fixed an issue with getting and setting Pipeline state that could cause native system memory leaks.
• Enhanced the OPC UA Events input to support custom event attributes that exist outside the default namespace using a N:attributeName syntax, where N in the namespace number.
• Fixed an issue where the REST Client connection could not connect to an endpoint that only supported TLSv1.3.
• Fixed an issue reading PI Event Frames that had null attribute values.
• Fixed an issue in the InfluxDB Connection with escaping spaces in measurement names and tags.