| Turbolinux Cluster LoadBalancer 10: User Guide | ||
|---|---|---|
| <<< Previous | Chapter 8. Architecture | Next >>> |
The clusterserverd daemon is the ringmaster that gets things going. It reads the configuration file to determine what needs to be done. It makes sure the network interfaces are configured correctly. It loads the kernel module if necessary, and configures it, using the /proc/net/cluster files. The daemon is started via the /etc/init.d/clusterserverd startup script. Like other startup scripts, you can specify the start, stop, and restart options to bring the service up or down. The script is normally called at boot time using the Sys V init scripts process.
Note that some of the configuration settings are used by the daemon itself, and some are used to configure the SpeedLink kernel module. For example, the ATM settings include ARP settings, delay settings, maximum number of entries in the tables, and the connection timeout value. The table sizes and connection timeout are used to configure the kernel module, while the rest of the settings are used by the daemon itself.
The kernel module does most of the traffic management work, deciding where to send each packet, and forwarding the packets. But it is the daemon that tells the SpeedLink module how to do its job. The daemon also monitors the servers and services and removes them from the kernel tables if they are found to be down or unresponsive.
CMC is provided as an additional tool to monitor the performance of the cluster. It does this by actively watching the /proc/net/cluster files and charting the progress. However, it can also be used to modify the daemon settings as well as the kernel module settings. It can tell the module to start, restart, or stop. It performs most of these tasks by communicating its wishes to the clusterserverd daemon. CMC actually runs as a daemon itself, named cmc. It has its own startup script, in /etc/init.d/cmcd.
The synchronization tools allow you to keep nodes in the cluster consistent. To run an effective cluster, you must maintain consistency. This includes not only the content, but the configuration settings as well.
By combining these various elements, the Turbolinux Cluster LoadBalancer 10 system is able to accomplish its goals. Multiple computers can be combined to provide enhanced speed, reliability, and scalability. The workload is distributed among several servers instead of concentrating all the work on one large server. The system can be set up to ensure that there is no single point of failure, greatly increasing the availability of your services.
Learning how everything fits together takes some time and effort, but in the end you will have a reliable and scalable system. After you have figured out how everything works, you will find that managing the cluster will come quite naturally. The combination of the technologies involved will help you meet your objectives while reducing costs and administration time.
| <<< Previous | Home | Next >>> |
| Cluster Management Console (CMC) | Up | Glossary |