The GUI provides a mechanism for you to test your computations, but ideally you want them to run automatically in the background when new data arrives. For this to happen, the following must be true:
- The computation must be enabled and assigned to a valid computation process
- That computation process must be running in the background
- One or more inputs for the computation must be arriving in the time-series database (TSDB)
Open the computation editor. Click the Computations tab along the top. Find the computation that is not working and double click. In the "Enabled for" field at the top:
- Make sure the checkbox is checked.
- Select a computation process (most likely "compproc") from the pull-down list.
- Commit the changes.
The default implementation is for a single computation process called simply "compproc". Your system administrator may have a more complex configuration.
From the Unix shell, type:
In the output, verify that the computation process you selected is running and that its 'heartbeat' is current (i.e. within the last minute or so).
Use your time series database editor (or import program) to add new input data for this computation.
You can check the computation processor log file (probably "compproc.log") to see any actions taken. It may be that your computation is being run but is failing. A common problem for Rating Table computations is that the rating table cannot be found.
The mechanism that causes a computation to be run on its inputs varies depending on the underlying database, but in all cases, background code within the database will add records to the CP_COMP_TASKLIST table for each computation input.