Re: More time spending with "delete pending"

Hello Michael,

06.07.2021 11:33, Michael Paquier wrote:
> On Sun, Mar 14, 2021 at 06:00:00PM +0300, Alexander Lakhin wrote:
>> I believe that the patch attached to [1] should fix this issue. The
>> patch still applies to master and makes the demotest (attached to [2])
>> pass. Also I've prepared a trivial patch that makes pgwin32_open() use
>> the original stat() function (as in the proposed change for _pgstat64()).
> Hmm. Knowing that _pgfstat64() has some special handling related to
> files pending for deletion, do we really need that on HEAD?
Yes, this fix is needed for HEAD (b9734c13) as the simple test attached
to [1] shows:
(You can put that script in src\test\modules\test_misc\t and perform
`vcregress taptest src\test\modules\test_misc`.)
t/001_delete_pending.pl ......... # Looks like your test exited with 2
before it could output anything.

and
src\test\modules\test_misc\tmp_check\log\regress_log_001_delete_pending
contains:
# Postmaster PID for node "node" is 1616
error running SQL: 'psql:<stdin>:1: ERROR:  could not stat file
"pg_wal/dummy": Permission denied'
while running 'psql -XAtq -d port=64889 host=127.0.0.1 dbname='postgres'
-f - -v ON_ERROR_STOP=1' with sql 'select count(*) > 0 as ok from
pg_ls_waldir();' at C:/src/postgresql/src/test/perl/PostgresNode.pm line
1771.

As Tom Lane noted above, the code added with bed90759f is dubious
(_NtQueryInformationFile() can not be used to handle the "delete
pending" state as CreateFile() returns INVALID_HANDLE_VALUE in this case.)
Probably that change should be reverted. Should I do it along with the
proposed fix?

[1]
https://www.postgresql.org/message-id/c3427edf-d7c0-ff57-90f6-b5de3bb62709%40gmail.com

Best regards,
Alexander