On Mon, Sep 13, 2021 at 8:39 PM Euler Taveira <eu...@eulerto.com> wrote:
>
> On Mon, Sep 13, 2021, at 10:09 AM, Amul Sul wrote:
>
> Yeah, added that test too. I triggered the restartpoint via a
> CHECKPOINT command in the attached version.
>
> +# archive_cleanup_command executed with every restart points
> +ok( !-f "$archive_cleanup_command_file",
> + 'archive_cleanup_command not executed yet');
>
> Why are you including a test whose result is known? Fresh cluster does
> not contain archive_cleanup_command.done or recovery_end_command.done.
>
Make sense, removed in the attached version.
> +# Checkpoint will trigger restart point on standby.
> +$standby3->safe_psql('postgres', q{CHECKPOINT});
> +ok(-f "$archive_cleanup_command_file",
> + 'archive_cleanup_command executed on checkpoint');
>
> Is this test reliable?
>
I think yes, it will be the same as you are suggesting a shutdown
which eventually performs a checkpoint. That checkpoint on the
recovery server performs the restart point instead. Still, there could
be a case where archive_cleanup_command execution could be skipped,
if there is no record replied after the previous restart point, which
could happen in case of shutdown as well. Hope that makes sense.
Regards,
Amul
diff --git a/src/test/recovery/t/020_archive_status.pl b/src/test/recovery/t/020_archive_status.pl
index cea65735a39..2277d09a306 100644
--- a/src/test/recovery/t/020_archive_status.pl
+++ b/src/test/recovery/t/020_archive_status.pl
@@ -8,7 +8,7 @@ use strict;
use warnings;
use PostgresNode;
use TestLib;
-use Test::More tests => 16;
+use Test::More tests => 18;
use Config;
my $primary = PostgresNode->new('primary');
@@ -234,3 +234,32 @@ ok( -f "$standby2_data/$segment_path_1_done"
&& -f "$standby2_data/$segment_path_2_done",
".done files created after archive success with archive_mode=always on standby"
);
+
+# Test archive_cleanup_command and recovery_end_command execution
+my $standby3 = PostgresNode->new('standby3');
+$standby3->init_from_backup($primary, 'backup', has_restoring => 1);
+my $standby3_data = $standby3->data_dir;
+my $archive_cleanup_command_file = "$standby3_data/archive_cleanup_command.done";
+my $recovery_end_command_file = "$standby3_data/recovery_end_command.done";
+$standby3->append_conf('postgresql.conf',
+ "archive_cleanup_command = 'echo archive_cleanuped > $archive_cleanup_command_file'");
+$standby3->append_conf('postgresql.conf',
+ "recovery_end_command = 'echo recovery_ended > $recovery_end_command_file'");
+
+$standby3->start;
+$primary_lsn = $primary->lsn('write');
+$standby2->poll_query_until('postgres',
+ qq{ SELECT pg_wal_lsn_diff(pg_last_wal_replay_lsn(), '$primary_lsn') >= 0 }
+) or die "Timed out while waiting for xlog replay on standby2";
+
+# archive_cleanup_command executed with every restart point and that can be
+# trigger using checkpoint
+$standby3->safe_psql('postgres', q{CHECKPOINT});
+ok(-f "$archive_cleanup_command_file",
+ 'archive_cleanup_command executed on checkpoint');
+
+# recovery_end_command_file executed only on recovery end which can happen on
+# promotion.
+$standby3->promote;
+ok(-f "$recovery_end_command_file",
+ 'recovery_end_command executed after promotion');
