Multiple nodes in a cluster provide resiliency against software error, and multiple nodes on separates machines add resiliency against hardware failure.
A typical and basic "safe-default" configuration for a Rhino SLEE cluster, is to use three machines, each hosting one node. Multiple nodes on separate machines for a cluster must be configured exactly the same (except for node ID). To do this, you can:
| Method | Description | Pros/Cons |
|---|---|---|
Install each node from the distribution .tar file, and be very careful to answer each question with exactly the same answer. |
✔ Does not require any special installer options or files to be copied during installation. ✔ Allows different nodes to be installed at different filesystem locations on each machine. ✘ Error-prone. ✘ Must copy keystores after installation. |
|
Install each node from the distribution .tar file, but create an "answers" file from the first installation and use it in the subsequent ones. |
✔ Avoids typos entering the configuration. ✘ Requires use of special installer options and copying a file during installation. ✘ Still requires keystores to be copied after installation. ✘ The Rhino directory must be in the same location on the filesystem on each machine, or configuration files must be edited. |
|
Install one node, then copy the entire base directory to each machine. This method is recommended for machines using the same configuration (program directory, JAVA_HOME, and so on). |
✔ Avoids typos. ✔ Saves having to find the keystores or answer files to copy. ✘ The Rhino directory must be in the same location on the filesystem on each machine, or configuration files must be edited. |
