Re: Determine parallel-safety of partition relations for Inserts - Mailing list pgsql-hackers

From Amit Kapila
Subject Re: Determine parallel-safety of partition relations for Inserts
Date
Msg-id CAA4eK1J3zJyoq4wtiWS9dH8XVMfhafq6V2y06CwxD8dr9+A-sg@mail.gmail.com
Whole thread Raw
In response to RE: Determine parallel-safety of partition relations for Inserts  ("Hou, Zhijie" <houzj.fnst@cn.fujitsu.com>)
Responses RE: Determine parallel-safety of partition relations for Inserts  ("Hou, Zhijie" <houzj.fnst@cn.fujitsu.com>)
List pgsql-hackers
On Thu, Jan 28, 2021 at 5:00 PM Hou, Zhijie <houzj.fnst@cn.fujitsu.com> wrote:
>
> > From: Amit Kapila <amit.kapila16@gmail.com>
> > > Good question. I think if we choose to have a separate parameter for
> > > DML, it can probably a boolean to just indicate whether to enable
> > > parallel DML for a specified table and use the parallel_workers
> > > specified in the table used in SELECT.
> >
> > Agreed.
>
> Hi
>
> I have an issue about the parameter for DML.
>
> If we define the parameter as a tableoption.
>
> For a partitioned table, if we set the parent table's parallel_dml=on,
> and set one of its partition parallel_dml=off, it seems we can not divide the plan for the separate table.
>
> For this case, should we just check the parent's parameter ?
>

I think so. IIUC, this means the Inserts where target table is parent
table, those will just check the option of the parent table and then
ignore the option value for child tables whereas we will consider
childrel's option for Inserts where target table is one of the child
table, right?

-- 
With Regards,
Amit Kapila.



pgsql-hackers by date:

Previous
From: Amul Sul
Date:
Subject: Re: [Patch] ALTER SYSTEM READ ONLY
Next
From: Peter Eisentraut
Date:
Subject: Re: BUG #16583: merge join on tables with different DB collation behind postgres_fdw fails