Calculation Job

The Calculation Job node offers the possibility to run distributed calculation jobs from within a DPS flow.
Configuration
Signals

Here it is possible to configure the input/output handling of the calculation.
For more details see: Signal Input and Signal Output.
Calculation

Here it is possible to specify the calculation type and upload the model.
- Calculation type
- Ebsilon
Select this type if the calculation is to run an EBSILON® Professional model. Details and the Extended Type can be found in the Ebsilon Calculation Type.
- Python
Select this type if you want the calculation to run a Python or IronPython model. For details and the Extended Type see Python Calculation Type.
Options

General Options
- Pass-through input data
If this option is enabled, all signals of the incoming data group are copied to the output data group. If the name of an output signal is identical to the name of an input signal, the output signal is used and the input is discarded. The input signal is even discarded if the calculation job does not contain the specified output signal or was manually deleted.
Default: enabled
- Delete processed job
If this option is enabled, the distributed calculation job will be deleted after successful processing.
Default: enabled
- Discard calculations with missing input values
If this option is enabled, no calculation will be performed if there are missing inputs.
Default: disabled
- Handle received jobs with execution outcome warning as error
If this option is enabled, jobs with the outcome “warning” will be treated as if they had the outcome “error”.
Default: disabled
- Include metadata signals on
out
port If this option is enabled, the metadata signals are written to the data group for the
out
port. See: metadata_signalsDefault: disabled
- Include metadata signals on
error
port If this option is enabled, the metadata signals are written to the data group for the
error
port. See: metadata_signalsDefault: enabled
- Create a debug archive for failed task items
If this feature is enabled, a debug archive is created for each failed task item. The debug archive can be accessed from the distributed calculation user interface.
Default: disabled
- Enqueue job in the specified execution queue (optional)
The execution queue of the distributed calculation service to use. If no queue is specified, the default setting is used.
- Set a calculation time limit for jobs (e.g. 5m or 00:05:00) (optional)
The maximum time after which a calculation must be completed (see Timespan ). If the calculation has not been completed or started after this time, it will be aborted.
Advanced Options
- Job creation mode
SingleTask AutoGrouped
Default: AutoGrouped
- Job parallel create limit
The maximum number of pending jobs created on the distributed calculation service. When the limit is reached, the node waits until a job is completed. Use
0
to disable the limit.Default: 0
- Discard enqueued datagroup messages when job creation limit is reached
If this option is enabled, the node will cancel the incoming calculation requests when the limit is reached (see Job parallel create limit).
Default: disabled
- Clear all stored jobs after worker reboot
If this option is enabled, all calculation jobs that have not been received will be removed on worker restart.
Default: disabled
- Clear all enqueued datagroup messages before start
If this option is enabled, all datagroups that are pending at the
in
port will be removed on start.Default: disabled
- Delete stored input datagroup of job
If this option is enabled, stored input datagroup of job will be deleted.
Default: enabled
Connection

- Calculation Connection
Select the distributed calculation connection to be used for this node (see Global Configuration).
Attention
Manual deletion of calculation jobs that have not yet been received by the Calculation Job
node can lead to unexpected behavior in the processing pipeline.
(e.g. the original data group id will be lost, because it is stored in the calculation job
and so the Flow Hook Receive
node will never catch the expected output message)