Schedule Refreshes on Tableau Online
You can schedule refresh tasks directly on Tableau Online for extracts of cloud-hosted data. This can include extracts of the following types of data:
Salesforce.com or Google Analytics.
Google BigQuery or Amazon Redshift, if you do not want to use a live connection.
SQL-based or other data hosted on cloud platforms, such as Amazon RDS, Microsoft Azure, or Google Cloud Platform, if you do not want to use a live connection, or if live connection isn’t supported for that data source.
For a list of supported data sources, see Allow Direct Connections to Data Hosted on a Cloud Platform.
- When a refresh is performed on extracts created in Tableau 10.4 and earlier (that is, a .tde extract), the extract is upgraded to a .hyper extract automatically. While there are many benefits of upgrading to a .hyper extract, you won't be able to open the extract with earlier versions of Tableau Desktop. For more information, see Extract Upgrade to .hyper Format.
- For information about refreshing extracts of Microsoft Excel, SQL Server or other data Tableau Online cannot reach directly, see Expand Data Freshness Options by Using Tableau Bridge.
When you’re signed in to Tableau Online, select Explore from the left navigation pane, and then, depending on the type of content you want to refresh, select All Workbooks or All Data Sources from the drop-down menu.
Select the check box for the workbook or data source you want to refresh, and then select Actions > Refresh Extracts.
In the Refresh Extracts dialog, select Schedule a Refresh, and complete the following steps:
Select the schedule you want.
If available, specify whether you want a full or incremental refresh.
A full refresh is performed by default. Incremental refresh is available only if you configured for it in Tableau Desktop before publishing the extract. For more information, see Refreshing Extracts in the Tableau Help.
Click Schedule Refreshes.
Scheduled refreshes run on Tableau Online whether or not you are signed in to your site. The refresh tasks are queued at the selected time and run once resources are available to execute the tasks.
Note: Refresh schedules may display a range of times, in addition to the specified refresh time. This is the naming convention used by Tableau to track refresh schedules and will not affect the execution of refresh tasks.
To ensure that long running refresh tasks don't take up all system resources and don't prevent refreshes of other extracts on your site, Tableau Online enforces a time limit, also known as a timeout limit, of 7200 seconds (120 minutes or two hours) for refresh tasks. The timeout limit is the longest allowable time for a single extract to complete a refresh before it's canceled. The timeout limit is not configurable.
Although reaching the timeout limit is not common, if you're working in an extract-heavy environment, there are few modifications you can make to your extracts to help avoid reaching the timeout limit.
Beginning with Tableau 10.5, the format of new extracts changed from .tde to .hyper. Along with a number of optimizations, refreshing large extracts in the .hyper format can be faster. If you're working with .tde extracts, consider upgrading your .tde extracts to the .hyper format. For more information, see Extract Upgrade to .hyper Format.
Consider configuring your extracts to be incrementally refreshed instead of fully refreshed each time a refresh task is performed.
By default, extracts are set up to fully refresh. Although a full refresh gives you an exact copy of the data, a full refresh can take a long time to complete. To reduce the time it takes to refresh an extract, consider setting up incremental refreshes of your extracts instead. For more information, see Configure an incremental extract refresh in the Tableau Help.
Note: You must set up the incremental refresh before publishing the extract to Tableau Online.
You can help refresh tasks complete faster by decreasing the size of your extracts.
There are two common methods for decreasing the size of extracts: hide all unused fields and use data source filters to remove unused columns.
Hide all unused fields: Fields that you hide in your data source are excluded when the extract is created. To hide fields while creating the extract, see Create an extract in the Tableau Help.
Add data source filters: You can remove unused columns in your extract by adding a data source filter. For more information, see Filter Data from Data Sources in the Tableau Help.
Note: You must hide all unused fields or add data source filters before publishing the extract to Tableau Online.
If possible, consider refreshing your extracts outside of Tableau Online.
Tableau Desktop: You can manually refresh published extracts from Tableau Desktop. For more information, see Refresh Published Extracts from Tableau Desktop.
Tableau Bridge: Depending on the data source, you can use Tableau Bridge to set up and refresh published extracts on a schedule. For more information, see Use Tableau Bridge to Keep Tableau Online Data Fresh.
Tableau Data Extract Command-Line utility: Depending on the data source, you can use the command-line utility that comes with Tableau Desktop to programmatically refresh published extracts. For more information, see Automate Extract Refresh Tasks from the Command Line.
Consider changing when the refresh tasks occur. For more information, see Manage Refresh Tasks.
Errors when refresh tasks reach timeout limit
If a refresh task reaches the timeout limit, you or others might see one of the errors listed below. If you published the extract or are the extract owner, you might see one of these errors in an email notification. If you're a site administrator, you might see one of these errors in the administrative view about Background Tasks for Extracts.
- The query time resource limit (7200 seconds) was exceeded.
- com.tableau.nativeapi.dll.TableauCancelException: Operation cancelled.
- The query time resource limit (8100 seconds) was exceeded.
To help resolve and avoid these errors in the future, see the section above, Staying within the timeout limit.