We’re pleased to announce the version 8.0.4 release of Tungsten Cluster and Tungsten Replicator, a maintenance release focused on stability, operational visibility, security hardening, and safer failover behavior. This update introduces stronger diagnostic capabilities, modernized third-party libraries, improved observability for replication activity, and important fixes across Cluster, Replicator, Connector, and Manager components.
Security & Platform Modernization
Tungsten v8.0.4 includes a major refresh of third-party libraries to address critical and high-severity CVEs and keep deployments aligned with current security best practices.
Third-party libraries such as ZooKeeper, Log4j, Jackson, Kafka clients, PostgreSQL JDBC, Guava, Netty, and Protobuf have all been upgraded to newer supported versions. Several outdated or abandoned dependencies have also been removed entirely, reducing operational risk and simplifying long-term maintenance.
These updates help strengthen security posture while improving compatibility with modern runtime environments.
Better Diagnostics & Operational Visibility
This release significantly improves diagnostic collection and troubleshooting workflows.
To further improve visibility into replication activity, Tungsten introduces a new thl stats command. Administrators can now gather statistics about THL contents, including DML and DDL activity grouped by date, schema, and table - making it easier to analyze replication patterns and troubleshoot workload behavior.
The tpm diag command has been enhanced to automatically capture MySQL configuration files referenced through !include and !includedir directives, even when those files require elevated permissions. Diagnostic packages now also detect and report Java heap dump (.hprof) files, helping administrators quickly identify memory-related failures and share relevant data with support teams.
Improved Stability & Safer Failovers
There have been several enhancements in v8.0.4 to improve reliability across the Tungsten stack.
Tungsten Manager now includes a new configuration option that prevents failover candidates from being promoted if they remain stuck in a SYNCHRONIZING state beyond a configured threshold. This closes an important edge case that could previously introduce risk during failover or switch operations.
Shunned nodes can now also be placed into archive mode, giving administrators more flexibility when managing unavailable or isolated nodes.
For environments running large systems, both Replicator and Connector now automatically generate heap dumps during out-of-memory events, making root cause analysis substantially easier.
Replicator & Connector Improvements
Tungsten Replicator logging performance has been optimized for situations where UPDATE or DELETE statements do not affect rows. Previously, verbose logging in these rare scenarios could negatively impact performance at scale. The system now logs concise messages by default while still exposing detailed information when DEBUG logging is enabled.
For Docker-based deployments, it is now possible to deploy a cluster-extractor topology using the standalone Replicator package, expanding flexibility for containerized replication workflows.
Tungsten Connector also introduces more flexible routing behavior by allowing the dataservice name to be specified at connection time. This provides greater control over shard routing and multi-service environments directly through the connection string.
Bug Fixes & Reliability Improvements
This release includes a broad set of fixes across installation, tooling, and runtime behavior.
Upgrades from Tungsten 6.x now properly handle SSL-related tungsten.ini modifications during tpm validate-update. Command-line utilities such as tpm purge-thl, tpm report, and THL/query replicator tools have been refined for more predictable behavior and cleaner output.
Multi-site environments also benefit from Connector improvements that reduce the chance of brief application disconnections during site recovery events. Manager stability has been enhanced to prevent infinite restart loops in rare cluster join scenarios.
Known Issue
Customers using MariaDB 11.4 or later with SSL enabled should be aware of a known issue related to the default ssl-verify-server-cert=true behavior introduced by MariaDB. While Tungsten installs successfully, some utility scripts may not function correctly.
As a workaround, adding ssl-verify-server-cert=false to the MySQL configuration resolves the issue until a future patch release addresses the behavior.
Upgrade Notes
Tungsten Cluster & Replicator v8.0.4 is recommended for all v8 users, particularly those seeking stronger diagnostics, improved failover protection, and updated security dependencies.
As always, we recommend reviewing the full release notes and validating upgrades in a staging environment before deploying to production.
Learn More
For complete technical details, please review the Tungsten Cluster and Tungsten Replicator v8.0.4 release notes.
If you have questions about upgrading, failover policies, diagnostics, or deployment best practices, please contact Continuent Support or request a demo.
Comments
Add new comment