Configuring the SavWareHA Monitor

smon

smon is a program that will run on the Standby System and monitor the Primary System. In the case of a Primary System failure, smon can be configured to automatically switch to fallback mode and will send a message to the administrator(s). In the case of a network failure, the Primary System will ‘turn off’ the slave (Standby System) side of the mirror without interrupting processing.

Please note that disruption of a TCP/IP link (not the ‘LLI’ link) will cause up to a 9 minute time out. This appears not to be configurable. Savoir Technology suggests that the ‘LLI’ method be used where network packet security is not an issue. SavWareHA release 2.03a or higher should be installed before using the ‘LLI’ method.

The configuration file for ‘smon’ is ‘/etc/sentinel.d/links’. This file is primarily accessed on the Standby System; however, it should be copied to the Primary System so that the file is identical on both systems. Each line describes a link between the Primary and Standby System. The contents of this file are as follows:

net {machine name [time-out]}

‘machine name’ is the name of the Primary System. If the machine name is omitted, the Primary System name is assumed.

‘time-out’ is the default number of seconds that smon will attempt to access the Primary System. To set this value, the “machine name” field must be filled in. As a shortcut, you can specify the “machine name” as ‘-‘ to default to the Primary System name.

[optional]
serial primarydev=standbydev

The serial link between the Standby and Primary Systems.

[optional]
manual

Always ‘true’; failure of smon to sense the Primary System will not cause a fallback. If this is found anywhere in the file, the system will never automatically fall back.

‘smon’ will only cause an automatic fallback if all links fail, but will never cause a fallback if the keyword ‘manual’ appears anywhere in the ‘links’ file on a line by itself.

Changes to /etc/sentinel.d/links will take effect automatically within 60 seconds without requiring a system reboot. Use the sentinel menu option “Display Monitor Status” to show the current configuration and status.

For example, the file might contain:

net pluto 8
serial tty2a=tty1a

Please note the there is a [TAB] after the word ‘serial’ in this example. This indicates that the Primary System is "pluto", time-out is set to 8 seconds, and that there is a serial link between tty1a (COM1:) on the Standby System to tty2a (COM2:) of the Primary System.

Note: The cable required to connect two DB-9 COM ports is as follows:

2   ---   3
3   ---   2
5   ---   5

Note: The cable required to connect two DB-25 COM ports is as follows:

2   ---   3
3   ---   2
7   ---   7

The links can be monitored manually using the sentinel command with the ‘Utilities:Display Monitor Status’ option. One line will be displayed for each link defined in the links file. The link configuration and status information will be displayed.

Serial links may also be established between ports on terminal servers. The ‘smon’ program will periodically test each link and report failures to the system administrator(s). This technique allows automatic unattended notification of terminal server failures