Cluster implementation has been updated in CUBE 7 to make it easier to set-up Cluster processing in your model and to manage/select/monitor your cluster resources before and during your model run.

Cluster Start/Stop

CUBE 7 has a new tool (Cluster pool manager) to manage the Cluster resources available in your network environment. Cluster pools (groups of machines) are then available to select when running your models. To learn more about the cluster pool manager, please refer to Setting up Cluster Pools

With the introduction of Cluster pool manager, users do not have to use Cluster start/stop commands in their models. These statements can be removed from your model scripts.

Intrastep

Cluster intrastep distribute statement specification has been updated and simplified in CUBE 7. Users do not have to manage the allocation of cluster node numbers and process ids.

CUBE 6: DistributeIntraStep ProcessID='Test' ProcessList=2-16

CUBE 7: DistributeIntrastep MaxProcesses = 15

The "COMMPATH" and "WEIGHTS" parameters have been depreciated.

Multistep

Cluster multistep distribute statement specification has been updated and simplified in CUBE 7. User’s do not have to specify and manage Cluster node number and process ids. This has been replaced with the ALIAS parameter.

• ALIAS – defines a name for the multistep block. This name can be used in the BARRIER statement to check for process completion.

CUBE 6: DistributeMultiStep ProcessID=’Test’ ProcessNum = 1

CUBE 7: DistributeMultiStep Alias = ‘AM Assignment’

The COMMPATH keyword has been depreciated.

Checking Multistep Process Completion (Wait4Files)

The Wait4Files statement used to check for process completion in a multistep block has been replaced with the BARRIER statement which uses the ALIAS name instead of the process name/process id information.

CUBE 6: Wait4Files Files=Test1.script.end, Test2.script.end CheckReturnCode = T

CUBE 7: BARRIER IDLIST=’AM_Assignment’, ‘PM_Assignment’ CheckReturnCode = T

For sample Scripts using the new CUBE Cluster see CUBE 7 Cluster Sample Scripts

The update to Cluster statements discussed above can be done through the text editor in CUBE or through an external program such as Notepad++ (https://notepad-plus-plus.org/). Notepad++ provides advanced search capabilities to quickly search for these statements across all the script files in your model.