Managing Background Jobs in Tableau Cloud
In Tableau Cloud, users can schedule extract refreshes, subscriptions or flows to run periodically. These scheduled items are referred to as Tasks. The Backgrounder process initiates unique instances of these tasks to run them at the scheduled time. The unique instances of the tasks that are initiated as a result are referred to as Jobs. Jobs are also created for runs that are initiated manually, by clicking the Run Now option in the web interface, programmatically through REST API, or tabcmd commands.
For example, an extract refresh task is created to run daily at 9 AM. This is an extract refresh task, and every day at 9 AM, a job will be created for the Backgrounder to run.
Running all these jobs can mean that Backgrounder uses a lot of resources at various times during the day. Using the Job Management feature, Site administrators can get more details on these jobs that happen in their Site, and take action on those jobs to better manage resource usage. .
The Run Now settings on the General settings page also allows you to manage your resources by either allowing or blocking users from running jobs manually. By default, this option is selected to allow users to run jobs manually. Untick the box to prevent users from running jobs manually.
The Jobs page which contains the information about jobs can be accessed by navigating to the Existing Tasks menu of the left navigation menu.
Note: Information about jobs can only be viewed by site administrators.
This topic describes how to view and understand the information displayed in the Jobs page.
Your Tableau Cloud site comes with capacity to support all your users’ analytic needs. A site’s capacity includes capacity for storage and tasks that need to be performed on the site for extracts, metrics, subscriptions and flows. The jobs page allows you to monitor the tasks for your Tableau Cloudsite. For more information on site capacity, see Tableau Cloud Site Capacity.
At the top of the page, there are high-level statistics for the number of Failed, Completed and Cancelled jobs within the past 24 hours. Applying filters do not change these values.
For each job generated, there is a Job ID, the status of that job, the priority, the type of task that the job was generated from, the current run time – if the job is in-progress, current queue time – if queued, as well as the average run time and average queue time.
Tableau records historical run times and queue times to compute the average run times and average queue times. Both average run times and average queue times are calculated as weighted averages using the following formula: ((current run time or queue time average x 4) + most recent run time or queue time)/5.
The Job ID can be useful when viewing jobs on Admin views . When you click on the Job ID, you will see more detailed information about the job, such as the Job LUID, the project name, the schedule, the content name, content owner, job creator
Note: Doing a Refresh Now from the Data Sources page will only show the LUID information in the Job Details dialog box.
There are several types of tasks:
Bridge Refresh: Includes full and incremental extract refreshes that use Cloud schedules. For more information, see About Bridge Refresh jobs.
Extracts: This includes extract creation, incremental extract refreshes and full extract refreshes. .
Subscriptions: Includes subscriptions for workbooks and views.
For more information, see Create a Subscription to a View or Workbook.
Flow: This includes scheduled flows and manual flow runs.
Encryption: Includes the following:
- Extract encryption and decryption
- Flow encryption and decryption
- Re-key extracts and flows
You can filter to see only certain jobs. The available filters are by Job Status type, Task Type and Time Range. For the Time Range filter, you can choose from past one to 24 hours, in four hour increments.
Extract refreshes, subscriptions and flow run jobs can be cancelled. You can only cancel one job at a time, and selecting multiple jobs at one time for cancellation is not supported.
When you cancel a job, an email with the time the job was cancelled, the affected content and the time the job ran before being cancelled is sent to the recipients that you select in the Cancel Job dialog box . In addition, you can add your customised notes to be included in the email.
If you do not select any recipients, the job will be cancelled, but no email will be sent.
To cancel a job, click on the ellipses next to the Job ID and use the dialog to cancel the job:
There are seven types of status that jobs can be in, and hovering over each status will display more relevant information.
Completed: This job shows as Completed successfully and you can see the time when the job completed in the tooltip that is displayed when you hover over the status.
In Progress: This job shows as In Progress. A time for how long the job has been running for is displayed in the tooltip when you hover over the status.
In Progress: This job is In Progress, but is running late. Tableau keeps track of the average run times for the same job, and if the current run time is longer than the average run time, then it is considered running late. Times for how much longer than average the job has been running and its average run time is provided in the tooltip that is displayed when you hover over the status.
Pending: This job is currently Pending, waiting to be run when there is available Backgrounder capacity. A time for how long the job has been in the queue for is provided in the tooltip that is displayed when you hover over the status.
Pending: This job is currently Pending, but is running late. Tableau keeps track of the average queue times for the same job, and if the current queue time is longer than the average queue time then it is considered running late. Times for how much longer than average the job has been queued is provided in the tooltip that is displayed when you hover over the status.
Cancelled: This job was Cancelled by a Site administrator. The time the job was cancelled and how long it ran for before cancellation is provided in the tooltip that is displayed when you hover over the status.
Failed: This job is showing as Failed. The time when the job failed, how long it ran for before it failed, and why the job has failed is provided in the tooltip that is displayed when you hover over the status.
Suspended: This job is showing as Failed with a pause icon. If the job fails 5 times consecutively, the job will be suspended. Suspended tasks are still available but Backgrounder will not create jobs for these tasks until they are resumed by the user.
Although Bridge Refresh jobs can generate the same statuses as other job types, Bridge Refresh jobs differ in the following ways:
Cancelling jobs: Bridge Refresh jobs can't be cancelled from the Jobs page. Instead, ensure that Bridge clients are configured with the appropriate timeout limit to prevent any misuse of refreshes. For more information, see Change the Bridge Client Settings
"Sent to Bridge" job status: A "Sent to Bridge" status indicates a completed Bridge Refresh job. A completed Bridge Refresh job means that the refresh job was successfully sent to a Bridge client in the pool. A completed Bridge Refresh job does not indicate whether the refresh itself completed successfully. If a refresh fails for whatever reason, the publisher (data source owner) is notified through both an account alert and a failure email alert. These alerts provide the publisher troubleshooting steps to help resolve the issue.
- Subscription and Bridge Refresh jobs: Subscription jobs can't be initiated by completed Bridge Refresh jobs. This is because a completed Bridge Refresh job only indicates whether the refresh job was successfully sent to a Bridge client in the pool and not whether the refresh job was completed successfully.
- Refresh jobs that originate from Bridge (legacy) schedules are not captured on the Jobs page. To monitor refresh jobs for Bridge (legacy) schedules, you can refer to the Bridge Extracts admin view. For more information, see Bridge Extracts.
- For troubleshooting Bridge errors you see on the Jobs page, see Troubleshoot pooling.