Superset 4.x FAQ: Handling Large Datasets, Custom Visualizations, and OAuth Setup

How large a dataset can Superset handle?

Superset works with very large databases because it is a lightweight layer on top of the underlying database or data engine; all processing happens in those systems. Superset only visualizes query results.

Performance depends on whether the database can execute queries and return results at a user‑acceptable speed. If Superset feels slow, benchmark and tune the data warehouse.

What compute specifications are required to run Superset?

The required specs depend on the number of users and their activity, not on data size. Community admins report that 8 GB RAM and 2 vCPUs are sufficient for a medium‑sized instance. Development tasks such as building images may need more resources.

Monitor resource usage and scale up or down as needed. Superset usage often spikes, for example when many participants load the same dashboard simultaneously.

The application metadata does not need a large database, although log files grow over time.

Can I join/query multiple tables at once?

In the Explore or Visualization UI a Superset SQLAlchemy datasource must be a single table or view.

Typical solution: create a table that contains all required fields via a periodic batch process.

Alternatively, create a view that abstracts arbitrary SQL queries as a virtual table, allowing joins and transformations. Performance then depends on the underlying database, and you should avoid excessive GROUP BY because Superset adds its own GROUP BY , which can duplicate work.

When using SQL Lab, you can write arbitrary SQL queries that join multiple tables as long as you have access.

How do I create my own visualizations?

Read the "Creating visualization plugins" documentation:

https://superset.apache.org/docs/contributing/howtos#creating-visualization-plugins

Can I upload and visualize CSV data?

Yes. Follow the guide to enable and use CSV uploads:

https://mp.weixin.qq.com/s?__biz=MzA4Mzc4NTE5MQ==∣=2692296860&idx=1&sn=95d0eb9afcfb9f580e9c016796c5a4c6&scene=21#wechat_redirect

Why do my queries time out?

For long‑running queries launched from SQL Lab, Superset allows up to six hours before Celery terminates them. Increase the limit in the configuration, e.g.: SQLLAB_ASYNC_TIME_LIMIT_SEC = 60 * 60 * 6 If a 504 Gateway Timeout occurs while loading a dashboard or exploring a slice, the gateway or proxy (e.g., Nginx) may not receive a timely response from Superset. Superset has a client‑side timeout (default 60 s). Raise the server timeout in superset_config.py:

SUPERSET_WEBSERVER_TIMEOUT = 60

Why are maps invisible in geospatial visualizations?

Register a free account on Mapbox.com, obtain an API key, and add it to the .env file under the MAPBOX_API_KEY variable:

https://www.mapbox.com

MAPBOX_API_KEY = "longstringofalphanumer1c"

How to limit dashboard timed refresh?

Add a timed_refresh_immune_slices key to the dashboard JSON metadata to exclude specific slices:

{
   "filter_immune_slices": [],
   "expanded_slices": {},
   "filter_immune_slice_fields": {},
   "timed_refresh_immune_slices": [324]
}

In this example, all slices except 324 will be refreshed automatically.

Disable staggered refresh by setting stagger_refresh to false and adjust the interval with stagger_time (milliseconds):

{
    "stagger_refresh": false,
    "stagger_time": 2500
}

What to do if the table schema changes?

Navigate to Data → Datasets , click the edit icon next to the changed dataset, and under the Columns tab click Sync columns from source . New columns are merged in the background; then adjust column settings and save.

Which database engines can be used as Superset's backend?

Superset’s metadata store (the OLTP backend) has been tested with MySQL, PostgreSQL, and SQLite. These are recommended for production. Other OLTP databases may work but are untested. Reported that Microsoft SQL Server cannot be used as a backend. Column‑store or non‑OLTP databases are unsuitable.

https://github.com/apache/superset/issues/18961

How to configure OAuth authentication and authorization?

See the Flask‑AppBuilder configuration example:

https://github.com/dpgaspar/Flask-AppBuilder/blob/master/examples/oauth/config.py

Is there a way to force a dashboard to use specific colors?

Add a label_colors key to the dashboard JSON metadata to map labels to color codes, e.g.:

{
    "label_colors": {
        "Girls": "#FF69B4",
        "Boys": "#ADD8E6"
    }
}

Does Superset support a given database engine?

A database must have a SQLAlchemy dialect and a DBAPI driver that follow Python standards. Limited‑SQL engines can work if they expose required functions (date functions, sub‑query support, logging, cursor handling, etc.). Custom connectors can be written; the only native example is the Druid connector.

https://docs.sqlalchemy.org/en/13/dialects/

https://www.python.org/dev/peps/pep-0249/

Does Superset provide a public API?

Yes. A public REST API is available and continues to expand. Documentation and Swagger UI are reachable via the links below.

https://superset.apache.org/docs/api/

Endpoints under /api/v1 were initially defined in SIP‑17 and are being extended.

https://github.com/apache/superset/issues/7259

Enable Swagger documentation in superset_config.py:

FAB_API_SWAGGER_UI = True

How to view usage statistics (e.g., monthly active users)?

Superset does not include a built‑in usage‑statistics feature. Query the logs table in the metadata database to extract activity data. An example analysis article:

https://engineering.hometogo.com/monitor-superset-usage-via-superset-c7f9fba79525

What does the "Hours Offset" field do in the Edit Dataset view?

The field lets you add or subtract a number of hours from a time column, useful for converting UTC to a local timezone.

Does Superset collect any telemetry data?

By default Superset uses Scarf to collect basic telemetry during installation and runtime. Scarf tracks npm installations and page views via the scarf-js package, strips PII, and provides aggregated statistics. Users can opt out by setting the SCARF_ANALYTICS environment variable to false or following Docker‑specific instructions.

https://about.scarf.sh/

https://docs.scarf.sh/gateway/

https://www.scarf.sh/package-sdks

https://docs.scarf.sh/gateway/#do-not-track

https://docs.scarf.sh/package-analytics/#as-a-user-of-a-package-using-scarf-js-how-can-i-opt-out-of-analytics

Does Superset have an archive panel or recycle bin to recover deleted assets?

No. Deleted dashboards, charts, datasets, or databases cannot be recovered through the UI, though discussions are ongoing.

https://github.com/apache/superset/discussions/18386

Regularly back up the metadata database. To restore, start a Superset instance with a backup copy of the metadata DB and use the UI export button or the CLI command superset export-dashboards. Import the resulting .zip into the current instance, or programmatically export assets as backups.

Series

https://mp.weixin.qq.com/s?__biz=MzA4Mzc4NTE5MQ==∣=2692296860&idx=1&sn=95d0eb9afcfb9f580e9c016796c5a4c6&scene=21#wechat_redirect

https://mp.weixin.qq.com/s?__biz=MzA4Mzc4NTE5MQ==∣=2692296810&idx=1&sn=ae777d5ca6a46d30ccec436ab77ece32&scene=21#wechat_redirect

https://mp.weixin.qq.com/s?__biz=MzA4Mzc4NTE5MQ==∣=2692296804&idx=1&sn=87b7725314598cbbfff6e99d0fa7784d&scene=21#wechat_redirect

https://mp.weixin.qq.com/s?__biz=MzA4Mzc4NTE5MQ==∣=2692296866&idx=1&sn=c5aad6e581226033ae71b35638a6c87b&scene=21#wechat_redirect