How to Load Multiple Resilience Clients

In the information on this page [] is used to delimit optional items on command lines, the [ and ] should not appear in actual commands. Similarly ... is used to indicate items that may be repeated.

If a machine is loaded a pure client its local databases are inaccessible to other machines. If a machine is required to access datbases on other machines, and to allow access to its databases to other machines then it should be loaded as a Combined Client Server Machine.

Pure Client machines machines that do not act as servers are loaded as follows:

On all platforms the −use command tail is specified one or more times and specifies the default set of servers used to access DP4 databases. This server list can be over-ridden on a per table or per database operation basis using the network resilience API. An alternative form of this option −mustuse n is available and designates a server as a "vital" machine as explained here.

See below for the use of the -nolocal option.

Resilience Command Tails

There are a number of resilience command tails, which control behaviour in the event of a network error. These are supplied to AUXDISTR (not the database manager as in local resilience). They are explained in the section on Network Error Handling. Without one of these command tails, programs will terminate if there is a network problem.

The -nolocal Option - Multiple Resilence without Local Database

Normally Multiple Resilience clients have a local database. If you want a client to use multiple remote servers but not to have a local database you include the −nolocal option on the command line to AUXDISTR. The client will ignore the flags indicating which tables are local and which are remote, and will treat all tables as remote.

Obviously such a system is not as resilient as one with a local database.