Nonstop Database, HiRDB Version 9 System Operation Guide
If the system manager unit shuts down or a communication error occurs between the system manager unit and other units, message logs cannot be output normally and the following problems might arise:
To avoid these problems, you should consider use of message log output dispersion.
For a HiRDB parallel server configuration, you can select one of the following message log output methods (message log output destinations):
Normally, message logs are output using method 1. If necessary, change to method 2. Method 2 is called message log output dispersion.
You use the pd_mlg_msg_log_unit operand to select the desired method. If this operand is omitted, the default is method 1.
The following figures show message output methods 1 and 2.
Figure 8-2 Normal message log output method (method 1)
Figure 8-3 Message log output method when message log output dispersion is used (method 2)
Table 8-1 Message log output methods shows the message log output methods; Table 8-2 Advantages and disadvantages of the message log output methods shows the advantages and disadvantages of each message log output method.
Table 8-1 Message log output methods
Method selected# | Condition | Message log output destination |
---|---|---|
Method 1 (manager) | During normal operation | Message logs are output to the message log files and to syslogfile of the server machine where the system manager is defined. |
When an error or communication error occurs at the manager unit | Message logs are output to syslogfile of each server machine. Only some message log information is output. Also, the correct message logs might not be output. | |
Method 2 (local) | During normal operation | Message logs are output to the message log files and to syslogfile of each server machine. |
When an error or communication error occurs at the manager unit |
#: The value in parentheses ( ) is the value specified in the pd_mlg_msg_log_unit operand.
Table 8-2 Advantages and disadvantages of the message log output methods
Method selected# | Advantage | Disadvantage |
---|---|---|
Method 1 (manager) | Because message logs can be centrally managed from the server machine of the system manager, the job of monitoring messages is simpler than when local is specified. | If the system manager unit shuts down or a communication error occurs between the system manager unit and other units, the correct message logs might not be output. |
Method 2 (local) | Correct message logs are output even if the system manager unit shuts down or a communication error occurs between the system manager unit and other units. | Because message logs are output to each server machine, the job of monitoring messages is more complicated than when manager is specified. |
#: The value in parentheses ( ) is the value specified in the pd_mlg_msg_log_unit operand.
The following shows the environment settings for using message log output dispersion.
All Rights Reserved. Copyright (C) 2011, 2015, Hitachi, Ltd.