CORBA 2.6.1 Administration Guide

among Location Service Daemon (LSD), Comm Server, and TCP/IP processes:
1. You use PATHCOM to define the number of Comm Servers.
2. You use the
Configuration Tool to define the relationships among the LSD, Comm Servers, TCP/IP processes, remote clients, and
application servers.
3. You can use PATHCOM to locate and resolve a NonStop CORBA system performance problem.
4. You can use the
cfgmgt tool and the nsdstart script to configure Parallel Library TCP/IP to increase capacity.
The use of underlying networking services (X25AM and LAN) by TCP/IP can also affect the performance of the NonStop CORBA runtime
environment. You use SCF to establish and monitor those relationships.
Caution:
NonStop CORBA does not use TS/MP to provide automatic load balancing for Comm Servers. Instead, the
relationships among clients and the Comm Servers are fixed by the LSD configuration. Thus, assumptions
about TS/MP load balancing and performance tuning will not always apply to managing the NonStop CORBA
runtime environment.
For example, a bottleneck could occur in a Comm Server even if other servers in the pool were idle; in this
case, you would improve performance by using the
cfgmgt to modify the assignment of clients to Comm Servers
rather than by using PATHCOM to add Comm Server processes.
Often a perceived problem with NonStop CORBA system performance is actually the result of application design and configuration, such as the
use of requests to stateful objects and the locations and design of objects that frequently exchange data.
NonStop CORBA Troubleshooting
The most common problems preventing successful operation of the NonStop CORBA system are the following:
Configuration errors
Inconsistencies in configuration between the NonStop CORBA system and another system with which it must communicate
Other interoperability problems, such as differing implementations of a CORBA-defined feature on different platforms
Transient networking failures
Resolving a problem that arises in the interaction between the NonStop CORBA system and some other ORB typically requires collaboration
with the remote administrator. In this case your best tool is a full, up-to-date set of the NonStop CORBA and SCF configuration files. However,
several automated tools can also help you troubleshoot your local NonStop CORBA system:
PATHCOM lets you verify the current configurations of NonStop CORBA runtime processes.
The NonStop CORBA error log is the primary source of information about NonStop CORBA execution errors. A variable in the NonStop
CORBA configuration file controls the error logging.
The NonStop CORBA trace facility lets you follow the flow of control in an application to discover when and where an error occurred.
Each component has a variable in the configuration file that controls the tracing for that component.
SCF lets you verify the configuration of networking resources, validate addressing information, discover recurrent errors, and obtain line
traces and protocol traces.
The Ptrace utility helps you to interpret SCF traces.
The Event Management Service (EMS), which has several interactive and programmatic interfaces, helps you capture and integrate the
error information reported by different processes including TS/MP, NonStop TM/MP, the operating system, and the networking services.
See the EMS Manual for descriptions of the EMS interfaces.
Chapter 3. Configuration and Management Using
the
Console
Chapter 5. Managing Application Processes