Store Results

A normal dataset runs its query every time a widget asks for data. That’s what keeps it live, but a heavy query (millions of order lines, several joins, aggregates over years) runs again for every widget on every page load.

Store Results keeps a stored copy of the dataset’s rows instead. A background job rebuilds the copy on a schedule, and widgets read the copy. They load fast and the server does the heavy work once per refresh, not once per viewer. The price: figures lag by up to the refresh interval.

When to use it

Use it when a dataset is slow and its figures don’t need to be up to the minute: monthly summaries, history over several years, management dashboards opened by many people.

Keep datasets live when:

• people act on the figures straight away, such as today’s orders or open tickets;
• the dataset is quick anyway;
• the SQL uses runtime parameters ({{date_from}}, {{date_to}}, {{as_at_date}}, {{int:…}}). Those datasets can’t be stored.

Widget filters and global filters keep working on a stored dataset: they are applied to the stored rows when a widget asks. This is the difference from dashboard optimization, which stores each widget’s unfiltered answer and is bypassed as soon as someone changes a filter. For heavy dashboards that people filter by date, Store Results is the better tool. See Optimization.

Turn it on

1. Go to Odoo Dashboards → Configurations → Advanced Configuration → DTS and open the dataset.

2. Under Performance, tick Store Results.

3. Choose how often to rebuild the copy under Refresh:

   • Every few minutes: set Refresh Every (minutes), 30 by default and at least 5.
   • Daily at a set time: set Refresh At (03:00 by default) and the Timezone it’s in, which defaults to yours.

4. Save.

123

1. Refresh Stored Results
2. Store Results
3. Stored Copy: the state of the copy

The first copy is built in the background, usually within a few minutes. Until it’s ready, widgets keep reading live data.

Watch the stored copy

The form shows the state of the copy:

Field	Meaning
Stored Copy	Queued (waiting for the background job), Refreshing, Up to Date or Failed.
Last Refreshed	When the copy widgets read now was finished.
Last Refresh Started	When the last attempt began.
Last Refresh Took (s)	How long the last build took.
Last Refresh Error	Why the last attempt failed. Shown only after a failure.

Refresh Stored Results in the header queues a rebuild now. It runs in the background, so the page doesn’t wait; reload the form to see the state change.

How refreshing works

• The new copy is built next to the old one and swapped in when it’s complete. Viewers never see an empty or half-built dataset; during a refresh they keep seeing the previous copy.
• A background job named Odoo Dashboards: refresh stored datasets runs every 5 minutes and rebuilds the datasets that are due, one at a time.
• Every few minutes counts from the last attempt, so a refresh that fails waits a full interval before trying again.
• Daily at a set time rebuilds at the first run after that time. A failed daily refresh is retried every hour.
• Date and datetime columns of the copy are indexed, so date filters stay fast.

Caution

Refreshing needs Odoo’s scheduled actions to run. On servers with several workers, the server’s time limit for scheduled jobs (limit_time_real_cron) must be longer than the build takes, or the refresh is stopped and marked Failed with a message saying so. Ask whoever runs your server to raise the limit, or make the query lighter.

Things to know

• Relative dates are frozen at refresh time. A Where Clause such as This month or Last 30 days, or CURRENT_DATE in the SQL, is worked out when the copy is built. With a daily refresh, “today” in the copy is the day of the last refresh.
• Changing the dataset drops the copy. Adding a column, editing the SQL and clicking Generate Dataset: the dataset serves live data again until the next copy is built, which is queued straight away.
• Turning it off switches the dataset straight back to live data and deletes the copy.
• Export CSV runs the dataset’s query live, not from the copy.
• Duplicating a dataset doesn’t copy the setting. Turn it on again on the copy.
• A stored dataset holds all its rows, so keep its SQL to the history you need.
• Datasets whose names map to the same view name can’t be stored: Another dataset uses the same view: generate this one before storing it. See Naming datasets.

Validation messages on save: A dataset that uses runtime parameters ({{as_at_date}}, {{date_from}}, {{int:…}}) is computed per request and can’t be stored., Stored results can be refreshed at most every 5 minutes., and The refresh time must be between 00:00 and 23:59.