Submitting Commands

A client can submit commands and keyins to host systems that are being monitored by the Operations Sentinel server. The host must be active and connected to the Operations Sentinel server.

Logon Requirements for a Managed UNIX System

If a managed system requires the user to log on, as with a UNIX system, then a user must be logged in to the system before a client can submit a command. Operations Sentinel does not check that a user has logged in, so a client can submit a command without detection of an error.

Alternatively, you can design your client to send the login command, or to respond to prompts at boot time.

Identifying the Host

Clients must identify the host to which the command will be sent. This is done by using the host name that was used when the host was configured with Operations Sentinel Console. If an unrecognizable host name is used, the command is discarded and an error is returned on the command acknowledge. You must inform our customers to begin using the Internet host name of hosts when creating a configuration with Operations Sentinel Console. If they do this, then client programs can also use the Internet host name as the command destination.

Submitting a Command Using SPDCommand Function

At the time the command is submitted, there is no way to know if the host or console is being actively monitored by the Operations Sentinel server, or if the host or console is active. The command is received by the Event Server, which attempts to forward it to the specified host or console. Once the attempt is made, a status is returned to the requesting client. The command acknowledgment callback specified with SPDCommand function is then invoked with this status.

Using Passback Data

The client must use the passback data of the SPDCommand function to correlate a command with its acknowledgment. A unique value can be associated with each individual command in this way.