8.8.1 Tuning the monitoring interval of the operation state of CTM domain managers and CTM daemons

A communication process is executed periodically between the multiple CTM domain managers present in a system, and the multiple CTM daemons within a CTM domain, to monitor the mutual operation state.

This subsection describes the concept of tuning the respective communication process intervals.

Organization of this subsection
(1) Tuning the monitoring interval of the operation state between CTM domain managers
(2) Tuning the monitoring interval of the operation state between CTM daemons

(1) Tuning the monitoring interval of the operation state between CTM domain managers

The CTM domain managers periodically exchange information about the CTM daemons present in the mutual hosts. Based on this information, the requests received in the local host are distributed properly in the CTM daemon of the other host.

When exchanging information, the mutual operation state of the CTM domain managers is also checked concurrently. If the CTM domain manager of the other host is not running, requests will not be distributed to that host. While exchanging information, if a CTM domain manager does not receive a response until the lapse of a time period derived by multiplying with a fixed coefficient of time, the other CTM domain manager is assumed to be in a stopped state.

The setup location for the interval of exchanging information differs depending on whether the other CTM domain manager is present within the same network segment, or in another network segment. The default value of the coefficient used to judge the operation state is 2. For example, when checking the operation (online) information of a CTM domain manager present within the same network segment of a system built using Management Server, if there is no response for a time period derived by multiplying 2 with the dispatch interval of the CTM domain configuration information, the CTM domain manager is assumed to be in a stopped state. If the dispatch interval of the CTM domain configuration information is 60 seconds, the CTM domain manager will be assumed to be in a stopped state if there is no response for 120 seconds.

You can change this coefficient value to optimize the communication process.

Determine an appropriate value to be specified in the coefficient, after considering the load of the processing generated as a result of communication. If necessary, determine the standard dispatch interval as well. If you reduce the coefficient, the interval will be shortened, and you can immediately detect if the other CTM domain manager is in stopped state. This will help in preventing the dispatch of requests to the host in which the CTM daemon has stopped running. However, if the interval becomes too short, the communication processing will increase, thus increasing the communication overload.

(2) Tuning the monitoring interval of the operation state between CTM daemons

The CTM daemons mutually distribute requests based on the information exchanged between the CTM domain managers.

If no response is received from the CTM daemon to which requests are distributed, the CTM daemon distributing the requests assumes that the other CTM daemon is in stopped state, and does not distribute any more requests.

By default, a CTM daemon waits for 180 seconds to receive a response. If no response is received within 180 seconds, it is assumed that the other CTM daemon is in stopped state. By changing this value, the unnecessary wait time can be shortened.

You set up an appropriate value after considering the size of the data to be sent as a request. By shortening the interval, any trouble in the other host can be detected promptly, and therefore, enabling you to disconnect from the system at a stage when the effect of the trouble is minimum. However, if the interval is too short, timeouts may occur while a large-sized data is being transferred.