8.4. Synchronization Tools

The synchronization tools provide coherency among the nodes in the cluster. We talked about how they are used in the previous chapter. In this section, we will explain how they are implemented.

The synchronization tools use Secure Shell, or SSH, to transfer files between systems. The source system is always the system that you are running the utility on. The other systems listed will receive the content. Note that there is no turning back -- once you have committed to sending changes, whatever is on the source system gets transferred to the others.

The actual file transfers are performed by the scp program. Like all the programs in the SSH suite, you will be required to type in a password the first time you connect to each system. Use the root password for each system. SSH will remember the connection settings, so you will not have to type in the password after the first successful connection. The password is stored in encrypted form, so you do not need to worry about the password being stored in clear text on the system. SSH does its work securely at all times.

The configuration synchronization program (tlclb_config_sync) is just a special case of content synchronization. It copies the files in the /etc/clusterserver directory so that all the systems are configured in exactly the same way. It also copies the license files, so that all the systems will be able to run.

The version of SSH that is installed along with Turbolinux 10 Server is openssh-3.9p1-3. If you have a version of SSH on your other cluster nodes, you can include them in the synchronization process, even if they are not running the Turbolinux 10 Server software. However, there are some known incompatibilities between different versions of SSH, so you may need to switch versions if you have any difficulties. Systems that do not have SSH installed should be removed from the list of servers in the synchronization programs. You will need to synchronize content on those systems by hand.

If you have more demanding content synchronization needs, you may want to look into a more robust solution. The rsync system is quite good at synchronizing data between several systems, and can figure out what needs to be synchronized from multiple sources. A secure version called ssync has also been implemented. As was mentioned in Chapter 2, the ultimate solution is to deploy a distributed file system to maintain consistent content between all the nodes, or to use some hardware shared storage solution.