You can schedule Jobs in advance and set them to execute on a regular cadence. These can be useful when you have a data source that is updated regularly.
-
Go to a project.
-
From the navigation pane, click Scheduled Jobs.
-
Click New Scheduled Job.
-
In the Create a Scheduled Job page, define the job:
- Scheduled Job Name
-
Enter the name of the job. The Jobs Dashboard lists each job by this name.
- File Name
-
Enter the name of the file to execute. Include any optional arguments to pass to the file.
- Hardware tier
-
Select the hardware tier used by the Job.
- Environment
-
Select the compute environment used by the Job.
- Data
-
Click to expand the section to see the Datasets configuration used by the Job.
-
Click Next. Then, define the Compute Cluster:
- Attach Compute Cluster
-
Use this option to provision and attach a compute cluster to the Job. The remainder of the configurations are explained in:
-
Click Next. Set up the Schedule:
- Use custom expression
-
Enter a custom Quartz CronTrigger expression. For example, if you wanted to run the job on the 5th minute of every day, enter the following: 0 5 * ? * *
- Repeat every
-
Set the frequency at which you want the Job to repeat.
- Run sequentially
-
Setting a Job to Run sequentially will cause the scheduler to always wait for the last Job it started to complete before starting the next one. For example, if you set up a scheduled Job to run once per hour, and one of the Jobs launched by the scheduler takes 90 minutes to complete, the next hourly Job will not start until the previous one has finished. Otherwise, multiple Jobs from this scheduler will be allowed to run simultaneously. The scheduler will not wait for the previous Job to finish if it’s still running. This mode should be used when your Job doesn’t depend on output from the previous Job.
-
Click Next. Set up Actions:
- Notify Emails
-
Enter a list of email addresses to notify when the Job completes.
- Update Model API
-
If a Model API has been publishing from the Project, the selected Model API will be republished after the Job has completed. Use this for retraining and updating a Model API regularly.