After the OUTPUT and NOTIFY operands have been specified on the SET AUTODIRECT command, they are used in the processing for automatic direction in the following ways.

SET AUTODIRECT (OUTPUT(WARN(NODE1.ANN)) NOTIFY(FAIL(NODE1.ANN)))
SET AUTOPWD    (OUTPUT(WARN(NODE1.ANN)) NOTIFY(FAIL(NODE1.ANN)))
SET AUTOAPPL   (OUTPUT(WARN(NODE1.ANN)) NOTIFY(FAIL(NODE1.ANN)))

SET AUTODIRECT (OUTPUT(WARN(NODE2.SAM)) NOTIFY(FAIL(NODE2.SAM)))
SET AUTOPWD    (OUTPUT(WARN(NODE2.SAM)) NOTIFY(FAIL(NODE2.SAM)))
SET AUTOAPPL   (OUTPUT(WARN(NODE2.SAM)) NOTIFY(FAIL(NODE2.SAM)))

1. When an update has been automatically directed to a remote node and execution is completed, the OUTPUT and NOTIFY settings are used on the node at which the automatically directed update executes. The OUTPUT and NOTIFY settings on the node at which the update originally executed are not relevant.

   For example, a user on NODE1 runs an application that performs an update, and the update is automatically directed to NODE2. On NODE2, the update runs with a return code of 8. SAM on NODE2 gets a NOTIFY message and the OUTPUT containing the error messages. Even though the update originally executed on NODE1, the OUTPUT and NOTIFY settings on NODE2 are used instead of the OUTPUT and NOTIFY settings on NODE1.

   The update has successfully executed on one node and is automatically directed to another node. The NOTIFY and OUTPUT settings on that second node determine who is notified of the update completion.

2. When an error occurs while attempting to automatically direct an update to a remote node, the OUTPUT and NOTIFY settings are used on the node at which the update originally executes.

   For example, a user on NODE1 runs an application that performs an update, and the update is to be automatically directed to NODE2. As the update is placed in the OUTMSG workspace data set (which contains work items to be sent to NODE2), the data set becomes full; therefore, the update is not automatically directed to NODE2. ANN on NODE1 gets a NOTIFY message and the OUTPUT containing the error messages.

   The update has successfully executed on one node and is to be automatically directed to another node. An error occurs before the request arrives at the other node. The NOTIFY and OUTPUT settings on the first node determine who is notified of the error. This is treated as an error case (like an update which executed with return code 8), so any user with the FAIL, WARN, or ALWAYS setting would be notified or would receive output.

Guidelines the use of OUTPUT and NOTIFY

To make sure that the appropriate users are notified or receive output, specify the same users on the OUTPUT and NOTIFY operands on the SET command issued on each node with automatic direction active.

The OUTPUT and NOTIFY lists should include users from at least 2 different nodes, if possible. This is important in the case of a workspace data set problem. For example, suppose a command executes successfully on NODE1 but cannot be sent to NODE2. If all the NOTIFY and OUTPUT users are also on NODE2, it is possible that RRSF might experience the same problem trying to notify the users as it experienced trying to send the command. By also specifying a user on NODE1 or by specifying one local user on each node to get the output, you ensure that RRSF can find someone to notify.