1.20. Tungsten Clustering 6.0.1 GA (30 May 2018)

Version End of Life. 31 July 2020

This is a bugfix release.

Known Issue

The following issues may affect the operation of Tungsten Cluster and should be taken into account when deploying or updating to this release.

  • Installation and Deployment

    • It was previously impossible to change from a non-SSL installation to an SSL installation using self-generated certificates if an INI style configuration was being used. This can now be achieved by using the following command-line:

      shell> tools/tpm update --replace-release --replace-jgroups-certificate --replace-tls-certificate

      Issues: CT-442

    • Previously the system had been configured to dump heap files by default when the system ran out of memory which was useful for debugging by the development team. This has now been disabled.

      Issues: CT-604

Improvements, new features and functionality

  • Installation and Deployment

    • The tpm diag command has been improved to include more information about the environment, including:

      • The output from the lsof command.

      • The output from the ps command.

      • The output from the show full processlist command within mysql.

      • Copies of all the .properties configuration files.

      • Copies of all the cluster configuration and .properties files.

      • Copies of all the my.cnf files, including directory configurations.

      • The output from the connector cluster-status command.

      • The output from all services in active/active clustering deployments.

      • Improvements to the clarity of some commands.

      • The INI files used by tpm (if using INI installs) are included.

      Issues: CT-530, CT-611, CT-615, CT-623

  • Tungsten Manager

    • The REASON FOR MAINTENANCE MODE message has been updated when a failover has occured to specifically indicate a failover rather than a switch.

      Issues: CT-624

Bug Fixes

  • Tungsten Manager

    • A script used internally by the manager to determine the status of replication, called mysql_checker_query.sql, had been identified as providing bad information under certain complex circumstances. The effects of the bad script could include out of memory failures. The script and query has been rewritten.

      Issues: CT-457

    • The first execution of ls within cctrl within active/active clusters could fail to provide the cluster status information at the top (world) level.

      Issues: CT-551

    • Performing a switch in a two-cluster active/active deployment could fail if the cross-site replicators were not accessible.

      Issues: CT-592

    • An error executing the query checker script would not get identified and trapped properly.

      Issues: CT-632

    • Within a running cluster, managers on different hosts with a composite cluster could show different cluster state information after a switch operation.

      Issues: CT-633, CT-634

    • The API has been updated to improve compatiblity with the Tungsten Dashboard.

      Issues: CT-639

Tungsten Clustering 6.0.1 Includes the following changes made in Tungsten Replicator 6.0.1

Release 6.0.1 is a bugfix release.

Behavior Changes

The following changes have been made to Tungsten Replicator and may affect existing scripts and integration tools. Any scripts or environment which make use of these tools should check and update for the new configuration:

  • Command-line Tools

    • The tungsten_set_position and tungsten_get_position commands have been deprecated and will be removed in the 6.1.0 release. These commands only worked with MySQL datasources. Use the dsctl command, which works with a much wider range of datasources.

      Issues: CT-517

Improvements, new features and functionality

  • Command-line Tools

    • The trepctl services has been updated to support the auto-refresh option using the -r command-line optionoption.

      Issues: CT-627

    • The trepctl has been updated with a new command, servicetable command. This outputs the status information for multiple services in a tabular format to make it easier to identify the state for multi-service replicators. For example:

      shell> trepctl servicetable
      Processing servicetable command...
      Service | Status | Role | MasterConnectUri | SeqNo | Latency
      -------------------- | ------------------------------ | ---------- | ------------------------------ | ---------- | ----------
      alpha | ONLINE | slave | thl://trfiltera:2112/ | 322 | 0.00
      beta | ONLINE | slave | thl://ubuntuheterosrc:2112/ | 12 | 4658.59
      Finished servicetable command...

      The command also supports the auto-refresh option, -r.

      Issues: CT-637

Bug Fixes

  • Installation and Deployment

    • Support for the GEOMETRY data type within MySQL 5.7 and above has been added. This provides full support for both extracting and applying of the datatype to MySQL.

      This change is not backwards compatible; when upgrading, you should upgrade Replicas first and then the Primary to ensure compatibility. Once you have extracted data with the GEOMETRY type into THL, the THL will no longer be compatible with any version of the replicator that does not support the GEOMETRY datatype.

      Issues: CT-403

    • When using Net::SSH within tpm, more detailed information about any specific failures or errors is now provided.

      Issues: CT-523

    • tpm would mistakenly report issues with JSON columns during installation which no longer applies as JSON support for MySQL 5.7 was added in 6.0.0.

      Issues: CT-635

  • Command-line Tools

    • The tungsten_provision_slave could hang within different scenarios, including being executed in the background, or part of a background script or cronjob. The script could also fail to restart MySQL correctly

      Issues: CT-319, CT-572

    • The trepctl status would fail badly if the service name did not exist in the configuration, or if multipl services were configured.

      Issues: CT-545, CT-593

    • When using tpm with the INI method, the command would search multiple locations for suitable INI files. This could lead to multiple definitions of the same service, which could in turn lead to duplication of the installation process and occasional failures. If multiple INI files are found, a warning is now produced to highlight the potential for failures.

      Issues: CT-626

    • When setting optimizeRowEvents back to false (it is enabled by default), the replicator could fail with IndexOutOfBound errors.

      Issues: CT-631

    • Using trepctl qs where the sequence number could be larger than an INT would cause an error.

      Issues: CT-642

  • Oracle Replication

    • The prepare_offboard_fetcher script could fail due to the use of command that may not exist on some platforms. Under some circumstances the script could also be installed as non-executable.

      Issues: CT-420, CT-421

  • Heterogeneous Replication

    • The templates for ddlscan for MySQL to Oracle do not escape field names correctly.

      Issues: CT-249

    • When replicating data into MongoDB, numeric values and date values would be represented in the target database as strings, not as their native values.

      Issues: CT-581, CT-582

    • The default partition method used when loading data through CSV files showed an incorrect example format. Previously it was advised to use:


      It should just show the data format:


      Issues: CT-607

    • The Javascript batch loader for Redshift could generate an error when loading the object used to derive information during loading.

      Issues: CT-620

    • The templates for ddlscan for Oracle to Redshift failed to handle the NUMBER type correctly.

      Issues: CT-621

  • Core Replicator

    • Optimizing deletes in row-based replication could delete the wrong rows if the pkey had not been enabled.

      Issues: CT-557

    • The included Drizzle driver would incorrectly assign values to prepared statements if the fields in the prepared statement included a question mark

      Issues: CT-608

    • During replication, the replictor could raise the java.util.ConcurrentModificationException error intermittently.


      This change is not backwards compatible; when upgrading, you should upgrade Replicas first and then the Primary to ensure compatibility with the metadata.

      Issues: CT-618

  • Filters

    • The truncatetext filter was not configurable within all topologies. The configuration has now been updated so that the filter can be used in MySQL and other database environments.

      Issues: CT-386