Re: Allow CLUSTER, VACUUM FULL and REINDEX to change tablespace on the fly

On Mon, Jan 25, 2021 at 05:07:29PM +0900, Michael Paquier wrote:
> On Fri, Jan 22, 2021 at 05:07:02PM +0300, Alexey Kondratov wrote:
> > I have updated patches accordingly and also simplified tablespaceOid checks
> > and assignment in the newly added SetRelTableSpace(). Result is attached as
> > two separate patches for an ease of review, but no objections to merge them
> > and apply at once if everything is fine.
...
> +SELECT relname FROM pg_class
> +WHERE reltablespace=(SELECT oid FROM pg_tablespace WHERE spcname='regress_tblspace');
> [...]
> +-- first, check a no-op case
> +REINDEX (TABLESPACE pg_default) INDEX regress_tblspace_test_tbl_idx;
> +REINDEX (TABLESPACE pg_default) TABLE regress_tblspace_test_tbl;
> Reindexing means that the relfilenodes are changed, so the tests
> should track the original and new relfilenodes and compare them, no?
> In short, this set of regression tests does not make sure that a
> REINDEX actually happens or not, and this may read as a reindex not
> happening at all for those tests. For single units, these could be
> saved in a variable and compared afterwards. create_index.sql does
> that a bit with REINDEX SCHEMA for a set of relations.

You might also check my "CLUSTER partitioned" patch for another way to do that.

https://www.postgresql.org/message-id/20210118183459.GJ8560%40telsasoft.com
https://www.postgresql.org/message-id/attachment/118126/v6-0002-Implement-CLUSTER-of-partitioned-table.patch

--
Justin