RE: Separate out FileSet from SharedFileSet (was Re: pgsql: pgstat: Bring up pgstat in BaseInit() to fix uninitialized use o)

On Tuesday, August 24, 2021 6:25 PM Dilip Kumar <dilipbalaut(at)gmail(dot)com> wrote:
> On Tue, Aug 24, 2021 at 12:26 PM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>
> > > I was also thinking about the same, does it make sense to name it
> > > just ""%s/%s%lu.%u.fileset"?
>
> Done
>
> > I think it is reasonable to use .fileset as proposed by you.
> >
> > Few other comments:
> Done

After applying the patch, I tested the impacted features
(parallel hashjoin/btree build) with different settings
(log_temp_files/maintenance_work_mem/parallel_leader_participation/workers)
, and the patch works well.

One thing I noticed is that when enable log_temp_files, the pathname in log
changed from "xxx.sharedfileset" to "xxx.fileset", and I also noticed some
blogs[1] reference the old path name. Although, I think it doesn't matter, just
share the information here in case someone has different opinions.

Some minor thing I noticed in the patch:

1)
it seems we can add "FileSet" to typedefs.list

2)
The commit message in 0002 used "shared fileset" which should be "fileset",
---
Instead of using a separate shared fileset for each xid, use one shared
fileset for whole lifetime of the worker...
---

[1] https://blog.dbi-services.com/about-temp_tablespaces-in-postgresql/

Best regards,
Hou zj