Re: TRUNCATE on foreign table - Mailing list pgsql-hackers

On 2021/04/09 11:05, Zhihong Yu wrote:
> 
> 
> On Thu, Apr 8, 2021 at 6:47 PM Bharath Rupireddy <bharath.rupireddyforpostgres@gmail.com
<mailto:bharath.rupireddyforpostgres@gmail.com>>wrote:
 
> 
>     On Thu, Apr 8, 2021 at 6:44 PM Fujii Masao <masao.fujii@oss.nttdata.com <mailto:masao.fujii@oss.nttdata.com>>
wrote:
>      > The followings are the open items and discussion points that I'm thinking of.
>      >
>      > 1. Currently the extra information (TRUNCATE_REL_CONTEXT_NORMAL, TRUNCATE_REL_CONTEXT_ONLY or
TRUNCATE_REL_CONTEXT_CASCADING)about how a foreign table was specified as the target to truncate in TRUNCATE command is
collectedand passed to FDW. Does this really need to be passed to FDW? Seems Stephen, Michael and I think that's
necessary.But Kaigai-san does not. I also think that TRUNCATE_REL_CONTEXT_CASCADING can be removed because there seems
nouse case for that maybe.
 
> 
>     I think we should remove the unused enums/macros, instead we could
>     mention a note of the extensibility of those enums/macros in the
>     comments section around the enum/macro definitions.

+1


> 
>      > 2. Currently when the same foreign table is specified multiple times in the command, the extra information
onlyfor the foreign table found first is collected. For example, when "TRUNCATE ft, ONLY ft" is executed,
TRUNCATE_REL_CONTEXT_NORMALis collected and _ONLY is ignored because "ft" is found first. Is this OK? Or we should
collectall, e.g., both _NORMAL and _ONLY should be collected in that example? I think that the current approach (i.e.,
collectthe extra info about table found first if the same table is specified multiple times) is good because even local
tablesare also treated the same way. But Kaigai-san does not.
 
> 
>     IMO, the foreign truncate command should be constructed by collecting
>     all the information i.e. "TRUNCATE ft, ONLY ft" and let the remote
>     server execute how it wants to execute. That will be consistent and no
>     extra logic is required to track the already seen foreign tables while
>     foreign table collection/foreign truncate command is being prepared on
>     the local server.

But isn't it difficult for remote server to determine how to execute? Please imagine the case where there are four
tablesas follows.
 

- regular table "remote_parent" in the remote server
- regular table "remote_child" inheriting "remote_parent" table in the remote server
- foreign table "local_parent" in the local server, accessing "remote_parent" table
- regular table "local_child" inheriting "local_parent" table in the local server

When "TRUNCATE ONLY local_parent, local_parent" is executed, local_child is not truncated because of ONLY clause. Then
ifwe collect all the information about context, both TRUNCATE_REL_CONTEXT_NORMAL and _ONLY are passed to FDW. In this
casehow should FDW determine whether to use ONLY when issuing TRUNCATE command to the remote server? Isn't it difficult
todo that? If FDW determines not to use ONLY because _NORMAL flag is passed, both remote_parent and remote_child tables
aretruncated. That is, though both local_child and remote_child are the inheriting tables, isn't it strange that only
theformer is ignored and the latter is truncated?
 


> 
>     I was thinking that the postgres throws error or warning for commands
>     such as truncate, vaccum, analyze when the same tables are specified,
>     but seems like that's not what it does.
> 
>      > 3. Currently postgres_fdw specifies ONLY clause in TRUNCATE command that it constructs. That is, if the
foreigntable is specified with ONLY, postgres_fdw also issues the TRUNCATE command for the corresponding remote table
withONLY to the remote server. Then only root table is truncated in remote server side, and the tables inheriting that
arenot truncated. Is this behavior desirable? Seems Michael and I think this behavior is OK. But Kaigai-san does not.
 
> 
>     I'm okay with the behaviour as it is consistent with what ONLY does to
>     local tables. Documenting this behaviour(if not done already) is a
>     better way I think.

+1


> 
>      > 4. Tab-completion for TRUNCATE should be updated so that also foreign tables are displayed.
> 
>     It will be good to have.

Patch attached.


> 
>     With Regards,
>     Bharath Rupireddy.
>     EnterpriseDB: http://www.enterprisedb.com <http://www.enterprisedb.com>
> 
> 
> w.r.t. point #1:
> bq. I think we should remove the unused enums/macros,
> 
> I agree. When there is more concrete use case which requires new enum, we can add enum whose meaning would be
clearer.

+1

Regards,


-- 
Fujii Masao
Advanced Computing Technology Center
Research and Development Headquarters
NTT DATA CORPORATION