Re: Add -k/--link option to pg_combinebackup

On Fri, 21 Feb 2025 at 03:50, Israel Barth Rubio <barthisrael(at)gmail(dot)com> wrote:
>
> There are some failures on compiler warnings and Windows jobs that
> seem unrelated with this patch.
>
> Besides that, there were failures when executing `010_links.pl` in the
> Linux autoconf jobs. As discussed in a side chat with Euler and Robert,
> and pointed out by Robert, Cirrus CI uses a very small
> `--with-segsize-blocks`. My tests in the V2 patch were assuming 1GB
> segments, so they were failing in Cirrus CI.
>
> The patch V3, which I'm attaching in this reply, implements some logic
> to handle any segment size.

Few comments:
1) The new file 010_links.pl added should be included to meson.build also:
'tests': [
't/010_pg_basebackup.pl',
't/011_in_place_tablespace.pl',
't/020_pg_receivewal.pl',
't/030_pg_recvlogical.pl',
't/040_pg_createsubscriber.pl',
],

2) Since it is a new file, "Copyright (c) 2021-2025" should be
"Copyright (c) 2025":
+++ b/src/bin/pg_combinebackup/t/010_links.pl
@@ -0,0 +1,212 @@
+# Copyright (c) 2021-2025, PostgreSQL Global Development Group
+#
+# This test aims to validate that hard links are created as expected in the
+# output directory, when running pg_combinebackup with --link mode.

3) Since it is a single statement, no need of braces here:
+ /* Warn about the possibility of compromising the backups,
when link mode */
+ if (opt.copy_method == COPY_METHOD_LINK)
+ {
+ pg_log_warning("--link mode was used; any
modifications to the output "
+ "directory may
destructively modify input directories");
+ }

Regards,
Vignesh