Should self hosted cluster support deletes?

[NickyHeuperman]

What type of bug is this?

Incorrect result

What subsystems are affected?

Query Engine

Minimal reproduce step

I'm trying to delete data from a table, I'm howver met with the following error:
Unsupported region request: Delete
Is there something I need to enable? Or are deletes simply not supported self hosted?

What did you expect to see?

Rows being deleted

What did you see instead?

Unsupported region request: Delete

What operating system did you use?

Docker

What version of GreptimeDB did you use?

0.14.3

Relevant log output and stack trace

[killme2008]

@NickyHeuperman Hi. The delete operation is supported https://docs.greptime.com/reference/sql/delete/

Are you trying to delete data from the metrics tables written by Prometheus remote write? Sorry, currently logical tables cannot be deleted—you need to delete from the physical tables instead, specifically from greptime_physical_table.

[killme2008]

To delete the prometheus_build_info metric table from greptime_physical_table, use this SQL statement:

delete from greptime_physical_table where __table_id in
  (select table_id from information_schema.tables where table_name='prometheus_build_info');

This query retrieves the table_id from information_schema.tables by table name and then deletes the corresponding data from greptime_physical_table using that __table_id (the logical table ID).

[NickyHeuperman]

Can I filter by time range here then? And/or tags?

[killme2008]

Of course you can.
For example, if you query the table_id from tables:

mysql> select table_id from information_schema.tables where table_name='prometheus_build_info';
+----------+
| table_id |
+----------+
|     1374 |
+----------+
1 row in set (0.03 sec)

Which is 1374, then you can delete data with timestamp or other columns:

delete from greptime_physical_table where greptime_timestamp > '2025-05-25 00:00:00' and  __table_id =1374;