User guide

A multi-process server is multi-threaded and may handle multiple concurrent clients
allocating each client its own subprocess for database access, thus allowing better
concurrency and availability.
The majority of the multi-process server operations are carried out in a client thread context
within the main server process, handing off control to the clients allocated subprocess only
when direct Oracle Rdb database operations are required. Each client has its own
OpenVMS subprocess, thus concurrency is improved, as the server does not need to
synchronize database operations.
By default, the allocated subprocess is terminated on client disconnect. Executors may also
be retained for re-use after client disconnect, see Prestarted Executors
for details.
A multi-process server is installed and invoked on each node from which you wish to serve
Oracle Rdb databases. Oracle Rdb must be already installed and running on these nodes.
The server communicates with the thin driver using Java sockets over TCP/IP with the
default Port ID 1701.
3.2.1 Starting a Multi-process Server
A multi-process server may be invoked by using the appropriate start statement within the
controller, as an Oracle SQL/Services JDBC dispatcher, or directly from the operating
system command line.
3.2.1.1 Starting a Multi-process Server from the Controller
A multi-process server may be started from the controller by referencing a multi-process
server definition in an XML-formatted configuration file. See Starting Servers
within
Oracle JDBC for Rdb Controller for more details.
Example
Given the following server section in the XML-formatted configuration file mycfg.xml:
<server
name="Mpserv1"
type="RdbThinSrvMP"
url="//localhost:1799/"
logfile="myLogs:serv1.log"
/>
the following command may be used to start this server from within the controller:
rdbthincontrol> start server Mpserv1
22