3.5.2 Creating a multi-polling environment on an agent
The configuration of distribution destinations is defined in the system configuration information. This means that the jobs destined for an agent are executed over a predetermined route. Also, an agent will only connect to the relay system that is defined as its connection destination in the system configuration information.
If you want to specify multiple distribution routes or for an agent to connect to multiple relay systems, you must create a multi-polling environment. You can assign an order of priority to the relay systems to which the agent connects. When an error occurs in a relay system in a multi-polling environment, the managing server can execute a job on the agent via another relay system. The agent can also connect to multiple relay systems according to the priority of each relay system at the time the connection is made.
- Organization of this subsection
(1) Setting up a multi-polling environment
You can assign an order of priority to higher systems in the Higher systems to be polled area under Settings to perform polling for multiple higher systems on the Communication settings page of the agent settings.
Add button
Adds a higher system as a connection target. You can add a maximum of eight higher systems. In the Addition of a higher system to be polled dialog box, specify the product type of the higher system and its host name or IP address.
Remove button
Removes the selected higher system from the list.
- button
-
Moves the selected higher system up one priority level.
- button
-
Moves the selected higher system down one priority level.
(2) Selecting the polling method in a multi-polling environment
An agent can find out whether there are any jobs addressed to it by checking with the higher relay system. This is called polling. The method of polling the relay system is governed by the settings in the Settings to perform polling for multiple higher systems area of the Communication settings page. There are two polling methods:
- Hot standby
-
In hot-standby mode, the agent only polls one relay system. Normally, the relay system with the highest priority in the Higher systems to be polled area is polled. If the agent cannot connect to that relay system, it polls the relay system with the second-highest priority. If this relay system cannot be reached, the agent polls the relay system with the next-highest priority. This continues until the agent successfully connects to a relay system, from which point it polls that system exclusively.
- Multiple hosts
-
In multiple hosts mode, the agent polls every relay system listed in the Higher systems to be polled area. Polling takes place in order from the relay system with the highest priority.
(3) Notes on multi-polling environments
Note the following when using the agent in a multi-polling environment:
-
For agents operating in a multi-polling environment, you can register ID groups only on the higher system with the highest priority.
-
In hot standby mode, if the relay system with the highest priority recovers from an error, you can have the agent poll that system again by shutting down the relay system to which the agent is connected. The agent will poll the recovered relay system when polling next occurs.
-
When you execute jobs on all routes in an environment that uses the hot standby method, the job on the route to the relay system to which the agent is connected is executed. However, the jobs on all other routes remain on the relay systems in Not running status. Therefore, you must use the managing server to periodically delete the jobs remaining on the relay systems and the completed jobs. If a large number of jobs are left on a relay system, the performance of the relay system might suffer.
-
When you execute jobs on all routes in an environment that uses the multiple hosts method, the agent receives the same job from each relay system it polls. In this scenario, we recommend that you clear the Replace existing package check box in the job installation conditions when you execute a remote installation job. If this check box is selected, a package is sent each time the job is executed, which might place an excessive load on the network. If the check box is cleared, any job that the agent has already received ends normally without the package being sent again. The job maintenance code in this scenario is 900090009000.