In a distributed system, you cannot search on the master indexes by default. Search requests are answered only by slave index servers, and not by the master index servers.
The default configuration has the following advantages:
The resources on the master index server do not need to be shared among indexing and searching processes.
A write variant and a read variant exist for each master index. If the master index server only carries out indexing, only the write variant has to be loaded to the main memory. If the master index server carries out indexing and searching, both variants have to be loaded to the main memory.
You can change the default configuration so that the master index servers are also used for searching. This makes sense in the following cases:
You can change the standard configuration in the following two ways:
Changing the setting for all new master indexes
Changing the setting for an existing master index