Delete data in Promscale

warning

There is no way to undo data deletion. If you need to restore your data, you must restore from backup.

Additionally, you can have data automatically deleted by setting retention policies. For more information about Promscale retention policies, see the .

Promscale exposes an HTTP API endpoint for deleting metric data by series, using labels. This works the same as the Prometheus HTTP API endpoint, except that you need to enable permissions for advanced users. Do this by setting the web-enable-admin-api flag to true.

note

The start and end timestamp options are not currently supported. The delete_series HTTP API endpoint lets you delete the metric series only across an entire time range.

This URL query uses a repeated label matcher argument that selects the metric series to delete. You need to provide at least one match[] argument:

Use the API endpoints:

  2. POST /delete_series
  4. PUT /delete_series
  2. curl -X POST -g http://localhost:9201/delete_series?match[]={job="prometheus", instance=~"prom.*"}

You can delete all data for a metric, using either an SQL query, or an HTTP API endpoint.

To delete all data for a metric from the database using a SQL query, use the prom_api.drop_metric(metric_name_to_be_dropped text) function. This is an administrative command, and works only when no Promscale instances are attached to the database.

For example, to delete the metric container_cpu_load_average_10s:

To delete the data for a metric from the database using the HTTP API, use the /delete_series endpoint and pass the metric name to be deleted as the matcher. This works differently to the SQL method shown, because it deletes all the individual data points for that metric, but leaves the metric itself.

To delete data in Promscale, you need to enable permissions for advanced users. Do this by setting the flag to true.

This URL query uses a repeated label matcher argument that selects the series to delete. You need to provide at least one match[] argument:

  2. POST /delete_series

For example, to delete all the data points for the metric container_cpu_load_average_10s using the /delete_series HTTP API:

You can delete metric data points based on time, using an SQL query. Any compressed data needs to be decompressed before performing the deletion. You can recompress the chunks later on, if necessary.

For example, to delete all data from the container_cpu_load_average_10s metric that is older than 10 hours, decompress all chunks related to the hypertable of the metric:

  2. SELECT decompress_chunk(show_chunks('prom_data.container_cpu_load_average_10s'));

Then, perform the deletion query:

If you want to delete a particular series from that metric only, you can use series_id=<ID> in the WHERE clause of the DELETE query.

Now, recompress the remaining data:

Delete trace data

note

To run this function: first stop the Promscale Connector, then connect to the database and run SELECT ps_trace.delete_all_traces();, finally start the Promscale Connector.