When a cluster of multiple servers is used it is important to be able to tell the members of the cluster apart. For this purpose each server in a cluster has a node name that must be unique within the cluster.
By default Yate uses the system's configured host name as a node name. This can be overriden by the nodename setting in section [general] of the main configuration file (usually yate.conf). A further highest priority override is provided by the -N command line parameter so the node name can be changed per run.
It is recommended that you choose node names that can be easily resolved to an IP address. By far the highest performance can be achieved by setting the node name to the unique part of the IP address of the node.
Suppose a node has a fully qualified hostname of mynode.domain.com and an IP address of 10.0.2.17 in a cluster of 10.0.x.y
- node-2.17 - good and fast, a single regexp can resolve that into 10.0.2.17
- mynode - good, can be resolved via DNS either directly or by adding the domain.com suffix
- somename - bad, there's no easy relation between the node name and its address