Open System Services Management and Operations Guide (G06.30+, H06.08+, J06.03+)

ln -s /G/system/ztcpip/networks networks
ln -s /G/system/ztcpip/protocol protocols
ln -s /G/system/ztcpip/services services
ln -s /G/system/ztcpip/hosts hosts
To use the full NonStop TCP/IPv6 addressing capabilities instead of the /etc/hosts file,
add the following NonStop operating system DEFINE to the TACLLOCL file that is executed
whenever a TACL session is started:
DEFINE =TCPIP^HOST^FILE, FILE $SYSTEM.ZTCPIP.IPNODES
If your users do not always login through TACL, follow the equivalent procedure for the
/etc/profile file to add the DEFINE to all OSS shells when they are started.
Note that OSS AF_INET or AF_INET6 sockets and the inetd process use almost the same
set of configuration files. NonStop TCP/IPv6 searches for host information can be controlled
by the environment variable TCPIP_RESOLVER_ORDER, as described in the environ(5)
reference page, either online or in the Open System Services System Calls Reference Manual.
This environment variable can be set in the .profile file on a user-by-user basis or in
/etc/profile for all shell processes that launch OSS sockets programs.
You can change the transport-provider process used for OSS AF_INET or AF_INET6 sockets
by specifying a NonStop operating system DEFINE. This ability is useful when your node runs
several copies of its TCP/IP processes or when you run more than one NonStop TCP/IP product
at the same time and want to distribute workload between them.
For example, the following DEFINE allows the $ZTC1 process to be used as an OSS AF_INET
transport-provider process:
add_define =TCPIP^PROCESS^NAME, FILE \$ZTC1
If the DEFINE is declared in the /etc/profile file, then all OSS AF_INET or AF_INET6
sockets applications that are started from an OSS shell prompt use the specified transport
provider process. If the DEFINE is declared in the .profile file of a specific user, you can
control workload distribution on the basis of user.
You must keep track of the process names or OSS process IDs (PIDs) that OSS sockets
applications use so that those processes running in a specific processor (and possibly all
processes used by the same application) can be stopped before you stop an OSS
transport-agent server.
You might want to run multiple TCP/IP processes for scalability or load-leveling. Then you can
assign OSS AF_INET or AF_INET6 sockets applications to a specific transport-provider
process so that the default transport-provider process for your node is not overloaded.
For systems running AF_UNIX Release 2 software (J06.05 and later J-series RVUs and H06.16
and later H-series RVUs), ensure that startup scripts do not include an entry to start $ZPLS if
the failure to start $ZPLS causes the script to be aborted and steps following the failing one
to be omitted. Attempts to start $ZPLS fail on these systems because the $ZPLS OSS sockets
local server has been replaced by the OSS local server 2 servers ($ZLSnn, where nn is the
processor number). OSS local server 2 servers are started automatically when the processors
are started.
When AF_UNIX Release 2 sockets are created, they are created in compatibility mode or in
portability mode. AF_UNIX Release 2 sockets can communicate with sockets in the same
mode. After a socket is created, the mode cannot be changed.
You can change the transport-provider process used for OSS AF_UNIX Release 2 sockets by
specifying the NonStop operating system MAP DEFINE =_AFUNIX_PROCESS_NAME to one
of these values:
$ZPLS
$ZAFN2
126 Managing Servers