| From: | senor <frio_cervesa(at)hotmail(dot)com> |
|---|---|
| To: | "pgsql-general(at)lists" <pgsql-general(at)lists(dot)postgresql(dot)org> |
| Subject: | sequence on daily log partitioned table |
| Date: | 2025-04-22 04:26:10 |
| Message-ID: | LV8P220MB09555DBF54A6F74131D92D88F7BB2@LV8P220MB0955.NAMP220.PROD.OUTLOOK.COM |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general |
Hi All,
I'm mainly interested in understanding how this works or why it can't, as opposed to just solving the problem. AI just told me this can't be done without a trigger but I'd like to confirm in case maybe I just asked the wrong question.
I want to have a partitioned log table receiving input from sensor equipment. The partitions would cover some time range like an hour or day. In all cases I want an ID column to default to a nextval from a sequence but starting over for each day. If I set the default on the partitioned table, I would need to alter it at midnight - not optimal. A default set on the partition hasn't worked and I'm not sure why this is a problem.
I've only ever used simplistic partitioning but this doesn't seem like an unusual need. I'm looking for the underlying strategy that prevents use of the sequence configured on the partition. A reference to something describing the decision flow would be fantastic.
I've been experimenting on version 13 out of convenience but if a newer version has relevant changes, I'm interested. I'm not interested in the discussions over whether the data volume is enough to justify partitioning.
Thanks in advance,
Senor
| From | Date | Subject | |
|---|---|---|---|
| Next Message | David G. Johnston | 2025-04-22 05:02:21 | Re: sequence on daily log partitioned table |
| Previous Message | Tom Lane | 2025-04-22 01:33:46 | Re: pg_get_serial_sequence not working for manually set seq |