Chapter 7. Tungsten Connector

Table of Contents

7.1. Connector/Application Basics
7.1.1. Connector Control Flow
7.2. Basic Connector Configuration
7.3. Clients and Deployment
7.3.1. Connection Pools
7.4. Routing Methods
7.4.1. Connector Routing Methods
7.4.2. Primary/Replica Selection
7.4.3. Connector Load Balancers
7.4.4. Specifying Required Latency
7.4.5. Setting Read Affinity and Direct Reads
7.4.6. Setting Read/Write Affinity in Composite Active/Active Environments
7.4.7. Setting Negative Affinity
7.4.8. Routing using embedded syntax in connect string
7.4.9. Connector Datasource Selection in Composite Clusters
7.4.10. Smartscale Routing
7.4.11. Direct Routing
7.4.12. SQL Routing
7.4.13. Host-based Routing
7.4.14. Port-based Routing
7.4.15. Read-only Routing
7.5. Using Bridge Mode
7.5.1. Enabling Bridge Mode
7.6. User Authentication
7.6.1. File Format
7.6.2. Direct Routing
7.6.3. Host Options
7.6.4. Updates
7.6.5. Generating Entries from a Script
7.6.6. Encrypting Data
7.6.7. Synchronizing Data
7.6.8. Limitations
7.6.9. Host-based Authentication
7.7. Testing Connectivity Via the Connector
7.7.1. Testing Connectivity in Bridge Mode
7.7.2. Testing Connectivity in Proxy Mode with No R/W Splitting Enabled
7.7.3. Testing Connectivity in Proxy Mode with R/W Splitting Enabled (SmartScale or @direct)
7.8. Connector Operational States
7.8.1. Connections During Automatic Failure/Failover
7.8.2. Connections During Manual Switch
7.8.3. Connections During Connection Failures
7.8.4. Other Errors
7.8.5. Connector Keepalive
7.8.6. Connector Change User as Ping
7.9. Connector/Manager Interface
7.10. Connector Command-Line Interface
7.11. Connector Inline Command Interface
7.11.1. Connector tungsten cluster status Command
7.11.2. Connector tungsten connection count Command
7.11.3. Connector tungsten connection status Command
7.11.4. Connector tungsten flush privileges Command
7.11.5. Connector tungsten gc Command
7.11.6. Connector tungsten help Command
7.11.7. Connector tungsten mem info Command
7.11.8. Connector tungsten show [full] processlist Command
7.11.9. Connector show slave status Command
7.11.10. Connector tungsten show variables Command
7.12. Advanced Configuration
7.12.1. Working with Proxy Protocol v1
7.12.2. Using Multiple Dataservices
7.12.3. Advanced Listeners
7.12.4. Connector Automatic Reconnect
7.12.5. Using the Connector with HA Proxy
7.12.6. Using Fall-Back Bridge Mode
7.12.7. Using the Max Connections Feature
7.12.8. Adjusting the Client Disconnect Delay During Manual Switch
7.12.9. Adjusting the Bridge Mode Forced Client Disconnect Timeout
7.12.10. Adjusting the Connnector Response to Resource Losses
7.12.11. Connector Logging Configuration
7.12.12. Connector Audit Logging
7.12.13. Connector SSL Advertisement Configuration
7.12.14. Connector IP Address Configuration
7.12.15. Deploying a Connector through Docker
7.13. Connector General Limitations

Tungsten Connector acts as a proxy service, sitting between client applications and datasources in order to balance the load and provide high availability (HA) support. The service works by accepting raw packets from clients, and then forwarding them to the datasource. The process is reversed when packets are sent back from the datasource and then redirected back to the clients.

Figure 7.1. Tungsten Connector Basic Architecture

Tungsten Connector Basic Architecture

In addition to this basic structure, Tungsten Connector also works with the other components of Tungsten Cluster to handle some specific scenarios and situations:

  • The connector works in harmony with the Tungsten Manager as part of Tungsten Cluster and enables the connector to redirect queries between known datasources within a given dataservice. For example, when the manager identifies a failed datasource, queries to that datasource are redirected to an alternative datasource without the application being aware of the change.

  • The connector works with the Tungsten Cluster configuration and a number of implied or explicit directives that enable the connector to redirect requests within different datasources within the network. For example, the connector can be configured to automatically forward write requests to a database to the active Primary within the dataservice and reads to active Replicas.

Throughout this process the connector is redirecting the network packets sent by application servers to the appropriate host. The contents and individual statements are not processed or accessed. At all times applications and clients using the connector do not need modification as to them it will appear as a MySQL server.

To start using the connector run the tpm connector command. This will open a connection with the MySQL CLI. See Section 7.3, “Clients and Deployment” for more detail on configuring your application to use the connector.


After installation the connector will only work with the --application-user and --application-password that were provided during installation. See Section 7.6, “User Authentication” if you need more information on adding users to

shell > tpm connector
Welcome to the MySQL monitor.  Commands end with ; or \g.
Your MySQL connection id is 93422
Server version: 5.5.34-log-tungsten MySQL Community Server (GPL)

Copyright (c) 2000, 2013, Oracle and/or its affiliates. All rights reserved.

Oracle is a registered trademark of Oracle Corporation and/or its
affiliates. Other names may be trademarks of their respective

Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.