The Unified CCE Service Control application is an interface into the Windows platform's service control manager, which starts and stops services.
Figure 1. Unified CCE service control
To start, stop, or cycle the processes in the CTI OS Server, use the appropriate tabs from the Unified CCE Service Control window. To set CTI OS to start automatically on Windows startup (the recommended method), select the service name and click the Automatic button.
When the CTI OS service starts, it launches processes listed in the following table. Some of these processes open and run in console windows. You can minimize, but not close these windows. Closing the console window in which a process is running forces a restart of that process.
Table 1 CTI OS system processes
Runs In Console Window
The main CTI OS Server process. This process manages all CTI OS objects and listens for and manages client connections.
The CTI OS tracing utility. This process uses the Unified ICM Event Management System (EMS) to trace server messages to local log files in EMS format.
The Unified ICM NodeManager (fault tolerance manager). Each Unified ICM service is started by NodeManager, and NodeManager restarts any abnormally terminated processes.
The Unified ICM NodeManagerManager (system fault tolerance). Each Unified ICM Node (e.g. CTI OS) starts up a NMM process to handle system-level faults. In the event of a unrecoverable system fault, NMM restarts the host computer.
CTI OS failover
The server processes are managed by a fault tolerance/recovery platform called NodeManager. NodeManager creates and monitors each process running as part of the CTI OS service, and automatically restarts abnormally terminated processes.
For CTI OS failover to work correctly, you must correctly configure the settings for the peripheral in the Peripheral Gateway (PG) Explorer. Perform the following steps to verify that the parameters are correct.
Run the PG explorer.
Click "+" to open the branch for the correct PG.
Select the peripheral.
Select the PERIPHERAL tab on the right.
In the edit field for the Configuration Parameters, enter "/LOAD 0" (without the quotes), if it is not already datafilled.
Cycle the PG.
"/LOAD 0" causes agents to be set to NotReady during a failover. By setting the agents to NotReady, calls are not routed to them and the failover is quicker and cleaner.
The default for ICM Version 4.6.1.x and earlier was "/LOAD 0". It was changed in Versions 4.6.2, 5.0, and 6.0 to "/LOAD 1". In Version 7.0, the default was changed back to "/LOAD 0". If you upgraded from 4.6.1.x to 4.6.2, 5.0, or 6.0, you must go into the PG explorer and manually set the parameter to "/LOAD 0" if you are using CTI OS. The default remains "/LOAD 0" in Release 9.0(1).
CTI OS 7.0 does not support "/LOAD 1". Therefore, if RejectIfAlreadyLoggedIn is set to 1, then the behavior of the system is as follows:
If RejectIfAlreadyLoggedIn flag is on, then no two agents can log in using the same agentID.
The agent desktop crashes while the agent is successfully logged in to the CTI OS Server and does not have calls. In this case, the supervisor must first log out the agent, and then the agent must re-launch the agent desktop, supply the correct password, and manually log in.
The agent desktop crashes while the agent is logged in to the CTI OS Server and has calls. In this case, the supervisor must first log out the agent, and then the agent must re-launch the agent desktop, supply the correct password, and manually log in.
Failover of CTI OS related components
CTI OS handles failover of related components as described in the following sections.
If an IP phone goes out of service, CTI OS sends an event to all soft phones associated with the IP phone that their IP phone is out of service. In addition, the affected softphones display the message "Offline." When the IP phone is back in service, agents must manually log in.
If a switch goes out of service, CTI OS sends an event to all softphones associated with the switch that the switch is offline. In addition, the affected softphones display the message "Offline." When the switch is back in service, agents must manually sign in.
Because the Peripheral Gateway (PG) is a fault-tolerant process pair, CTI OS is not affected if the PG merely switches active sides. If the PG goes offline, CTI OS sends an "Offline" message to each softphone client.
CTI Server failure
On a CTI Server failure, the CTI OS Server usually reconnects almost immediately to the redundant CTI Server. If reconnection to the redundant CTI Server is not possible, the CTI OS Server sends a failure response to any requests made to the CTI Server.
In addition, CTI OS sends an event message to all softphone clients. On receipt of this message, the softphone clients display an "Offline" message.
When the CTI Server comes back online, CTI OS performs a snapshot of all agents, devices, and calls to reestablish state information.
CTI OS Server failure
On a CTI OS Server failure, CTI OS disconnects all softphones from the failed CTI OS Server. These softphones attempt to reconnect automatically to another CTI OS Server; if reconnection is not possible, CTI OS sends an event message to all softphone clients. On receipt of this message, the softphone clients display an "Offline" message.
NodeManager restarts the CTI OS Server. When the CTI OS Server process comes back online, CTI OS performs a snapshot of all agents, devices, and calls to reestablish state information.